Analyzing ports efficiency in the ASEAN and the V.I.S.T.A. by Data Envelopment Analysis (DEA) Technique Lirn, Taih-Cherng

Size: px
Start display at page:

Download "Analyzing ports efficiency in the ASEAN and the V.I.S.T.A. by Data Envelopment Analysis (DEA) Technique Lirn, Taih-Cherng"

Transcription

1 Analyzing ports efficiency in the ASEAN and the V.I.S.T.A. by Data Envelopment Analysis (DEA) Technique Lirn, Taih-Cherng Guo, Jiunn-Liang Assistant Professor Assistant Professor Department of Shipping & Department of Merchant Marine Transportation Management National Taiwan Ocean University National Taiwan Ocean University 2, Peining Road, Keelung City 2, Peining Road, Keelung City Taiwan Taiwan Fax: Fax: Li, Chuan-Chen Research Associate Department of Logistics Management, National Kaohsiung First University of Science and Technology 1,University Road, Yuancha, Kaohsiung County, Taiwan Fax: acedinos@hotmail.com Abstract: Emerging markets such as China and India have risen abruptly in recent years, with other newly developing countries followed closely behind. This paper employs the data envelopment analysis technique to estimate the efficiency of the major ports in the V.I.S.T.A. and the ASEAN countries. Research results indicated that the largest container port in Singapore, South Africa and Turkey were efficient ports in CCR and BCC model. Using the CCR model (self-assessment) and the CCR cross efficiency model (relative-assessment), it had been found that the major ports in Singapore and South Africa were situated in Group I which was labeled as benchmark ports. Whereas the operational efficiency of the other VISTA ports were situated in Group Ⅲ, it shows lots of ports in emerging markets have not paid much attention on the operational efficiency of their ports.. Key Words: Container Ports, Emerging Markets, Data Envelopment Analysis 1. INTRODUCTION In market economy, enterprises have striven to reduce their cost to increase profits. Thus, they began to engage in international investment and trade, and result in the phenomenon of globalization. Globalization has a very significant impact on the world economics. In the same time, the raw material prices are raised because of the rapid economic development and strong demand in China. It also results in the raw material in short of supply globally. For those developing countries with abundant materials and resources, broad domestic market and high political stability, globalization helps their economics develop rapidly. These developing countries are coined as emerging markets. Goldman Sachs proposed a new term for four emerging economics, BRICs, which has attracted the global attention in The economics development of the BRICs countries is developed according to Goldman Sachs s forecast subsequently. In 2005, Goldman Sachs has 1

2 brought another issue, the Next Eleven, which also has attracted many public interests. After BRICs and the Next Eleven, the Japanese BRICs Economics Research Institute (JBERI) has proposed another new term for emerging markets, V.I.S.T.A. economics, which includes Vietnam, Indonesia, South Africa, Turkey and Argentina respectively. The institute believed there are five factors help the rapid economic development in BRICs countries, sufficient natural resources, adequate labor force, friendly foreign investment policy, high political and economic stability, and the strong spending power of their middle class. However, not many countries in the world could have these five strengths at the same time. If a country can have any four of the above-mentioned strengths, it will also have a very promising economic development. The term V.I.S.T.A. is coined to indicate five economics that have any four of the five economic strengths in It is said the GDP in V.I.S.T.A. will overtake G7 within the next 50 years, and the V.I.S.T.A. GDP will grow up 28 times to $268 trillion. This argument has received broad discussion among economists, enterprises and investors immediately. In the other parts of the world, some emerging markets in the Southeast Asia also have rich natural resources, advantageous geographic location and sufficient labor force which attracted many scholars research attention. These emerging economics were called Association of South East Asian Nation (ASEAN). ASEAN were founded in It include Indonesia, Malaysia, Singapore, Thailand, Philippine, Myanmar, Vietnam, Cambodia, Laos and Brunei. On January , ASEAN members have signed an agreement, Framework Agreement on Comprehensive Economic Co-Operation between the Peoples Republic of China and the Association of South East Asian Nations. It is estimated under the ASEAN-China Free Trade Agreement (ACFTA) framework, China and Southeast Asia will establish the world's biggest free trade area (FTA), liberalizing billions of dollars in goods and investments covering a market of 1.7 billion consumers and the total trade amount between its members will reach US$1.2 trillion. In terms of trade volume, ACFTA will become the third largest free trade region next to EU and NAFTA. In 2011, Japan and Korea will also negotiate with ASEAN members about free trade terms. It implies that the ASEAN will have very far-reaching influences on the global economy in the future. In general, a country will import and export most of its cargoes through its sea ports. Wu and Lin (2008) indicate that the port is an important link in a logistics chain and the port efficiency has a significant influence on a country s competitiveness. A high efficient port will reduce export levy, improve its commodities competitiveness in the international market. Thus, high port efficiency will help a nation s economic development, this is especially important for an emerging economics. As the extant researches on competitiveness of the emerging markets are scarce, researches on port efficiency in these markets are especially limited. Thus, this research looks into the major ports in V.I.S.T.A. and ASEAN member countries and analyzes their operating efficiency. There are very limited literatures available for port performance researches in the developing economics, and this research intends to fill this knowledge gap. This study firstly reviews the port efficiency and DEA model literatures, and explains how to choose adequate input variables and output variables to build up the DEA model and to carry out the port performance evaluation. cross-efficiency technique suggested by Doyle & Green (1994) was for the first time being employed to carry out targeted ports performance assessment on the VIST + ASEAN nations. Finally, the authors propose the conclusions and suggestions about how to improve the sea port operating efficiency and competitiveness in emerging markets. 2

3 2. LITERATURE REVIEW ON PORT PERFORMANCE AND DATA ENVELOPMENT ANALYSIS The data envelopment analysis (DEA) is one of the most common performance evaluation techniques employed by academicians in the management discipline. Roll and Hayuth (1993) are advocated to use DEA-CCR model as a research method to evaluate and improve ports efficiency. They use three input variables and four output variables to measure performance efficiency among 20 ports. But Roll and Hayuth s study was a DEA theoretical exploration and they use the hypothetical port data, not the real port data, in their research. Valentine and Gray (2001) applied DEA-CCR model to evaluate the top 30 container ports in the world, they used two input variables and two output variables to exam and explore the ships performance, the special form of ownership and organizational structure. Kaisar et al. (2006) used three input variables and one output variable to inspect 20 American ports efficiency. And the result revealed that container ports which possess railroad transportation networks will have higher operating efficiency than the others that do not possess one. Many scholars also discuss the applications of the basic model and modified models of the DEA technique. The latter model includes DEA-BCC, Additive model and Window analysis. These models are all used to measure the relative efficiency in different seaports and container berths. Martinez-Budria et al. (1999) used DEA-BCC model to measure 26 Spanish ports, they adopted 3 input variables and 2 output variables to compare the panel data from 1993 to They classify the sample seaports into three clusters before analyzing their operating efficiency. Tongzon (2001) chose DEA-CCR and Additive model to evaluate the operating efficiency in Australian seaports and other nations seaports. Six input variables and two output variables are used to measure the seaports performance. Their research result reveals that the DEA- CCR model was less efficient than Additive model, and concludes that Melbourne, Rotterdam, Yokohama and Osaka were comparatively inefficient ports. Itoh (2002) evaluated eight Japanese container ports efficiency (Port of Tokyo, Yokohama, Shimizu, Nagoya, Yokkaichi, Osaka, Kobe and Kitakyushu) between 1990 and Three different DEA models are used to evaluate the eight ports performance, namely, DEA-CCR, DEA-BCC and Window analysis. Itoh (2002) has adopted four input variables and one output variable to evaluate the Japanese container ports performance. In DEA-CCR model, Itoh (2002) found Port of Tokyo has maintained high efficiency level. In DEA-BCC model, it is suggested Yokkaichi and Shimizu should develop a business model of small-scale operation. Cullinane et al. (2005) also use DEA-CCR technique and DEA-BCC technique to rank the operation efficiency among the top 30 container ports in the world. They adopted 5 input variables and 1 output variable to measure the ports efficiency and inspect the relationship between port privatization and port operating efficiency. Barros and Athanassiou (2004) employ DEA-CCR technique and DEA-BCC technique to measure the ports efficiency in two Greece ports and four Portugal ports. Barros (2006) employed DEA-CCR, DEA-BCC, Cross-Efficiency and Super-Efficiency techniques to analyze the technical efficiency in Italian ports during He used a two-stage framework to explore 24 Italian ports efficiency. In the first stage, the DEA technique was used to calculate technical and scaled efficiency. In the second stage, the author used Mann-Whitney U-test to do hypothesis test. He used Cobb-Douglas productive function 3

4 to measure the productivity in Italian ports, and the input and output variables were chosen from macro economics discipline. The output variables chosen were liquid bulk cargo quantity, dry bulk cargo throughput, number of ships called, passengers embarked, containers throughput in TEUs, non-standard containers cargo throughput in weight and port revenues. Number of staffs, capital invested and operational costs were treated as input variables. The result revealed that most of the Italian ports had great administration. Some conclusions could be made by using the Mann-Whitney Z-test. Firstly, the larger ports had higher efficiency than smaller ports. Secondly, the containerized ports had higher efficiency score than low degree of containerized ports. Lastly, the ports with small number of employees were more efficient than the ports with large number of employees. Yip, Sun and Liu (2010) used a panel data on container terminal operators to study the container port operators production efficiency, quay superstructure, yard superstructures, port characteristics, and operator background were used as the four input variables; the terminal throughput is used as the sole output variable. Yip et al. found ports with more terminals will reduce their efficiency, and port operators with deeper water terminals are more efficient. Lin and Tseng (2007) employed five types of DEA models to evaluate the performance of major ports located in Asia-Pacific region. The operational efficiency of the ports in Hong Kong, Singapore, Pusan, Keelung, Kaohsiung, Tokyo, Yokohama, Kobe, Osaka and Nagoya are analyzed. Size of quayside container yard, total length of quays in container berths, and number of deep-water berth were selected as input variables. Number of vessel arrivals at port and containers throughput is treated as output variables. Most of the above-mentioned DEA literatures are focused on reviewing the ports efficiency in a single country, among different countries, among different regions, and in the top 100 ports in the world. Cullinane and Wang (2006) employed DEA-CCR and DEA-BCC model to measure the efficiency of 55 container ports and 69 terminals in Europe. They used the land size of a terminal, total length of container berths and number of equipments in the ports as input variables, and used container throughput (in TEUs) as outputs variable to measure ports efficiency. Dowd and Leachine (1990) have pointed out that the productivity in container ports/ terminals were dependent on the efficiently use of labor, land and equipments. Summarizing the above-mentioned literatures, the authors could conclude that each of the input variables and output variables could be classified into three major categories as follows. 1. Input variables should include labor, land/capital/equipment and other productivity input variables. (1) The labor variable The inputs of labor are the worker in the quayside, number of employees in ports, number of employees in terminals, and the expense of labor. Roll and Hayuth (1993) proposed that the number of workers in the quay was treated as a input variable, and Tongzon (2001) used the number of the employee in ports to replace the variable. González (2007) also used the number of the employee in port as input variable. Itoh (2002) found there was lack of the data which were include the number of the work in quay and the number of the employee in port, so he chosen the employee of the container terminal to replace these variables. Martinez- Burdia et al.(1999)and Liu (1995) used the expense of labor as input variables. M. Kim and Sachish (1986) chosen the market share of labor. Cullinane and Song (2003) advocated the number of employee in port and the workers in quay. Barros and Athanassiou (2004) also used the number of workers, but they didn t illustrate this variable. However, the number of working forces might have a highly negative correlation with the ports output. Low 4

5 automisation and mechanisation ports usually have higher number of employees and low output. Thus the number of work forces in the port is not considered as an appropriate input variable. (2) The land/capital/equipment variables In port industry, the proprietor invests huge capital in purchasing or renting land, loading/unloading equipment and infrastructure. We could find the indicator about land from the literature we reviewed. The total area of quay is used mostly by previous scholars. The indicators which are most used in loading/unloading are gantry crane, yard crane and straddle carrier. The indicators about infrastructure which were common are the number of container ships, the length of quay, the total length of berth and the length of berth. Liu (1995) use book value of fixed assets as input variable. Cullinane and Song (2003) used fixed assets in container yard and loading/unloading equipment as input variables. (3) Other input variables Tongzon (2001) used the number of tugs and the ship waiting time as input variables. Martinez-Budria et al. (1999) employed the discount of expense and other expense. Roll and Hayuth (1993) used the singularity of goods. 2. Output variables: We could separate into the actual amount of productivity, service level and other variables. As follow: (1) The actual production variable From literature review, it could find that the number of TEUs, the weight of goods and the number of ship come to port are the most used in evaluating the efficiency of operation. Most scholars major in using the number of TEUs or the weight of goods as the indicator of output variables. (2) The service level variable Roll and Hayuth (1993) chosen service level and customer satisfaction as output variables. Tongzon (2001) used ship working rate as an indicator, he calculated the service quantity in ports by measuring how many containers loading/unloading per ship per hour. (3) Other output variables: González (2007) advocated output variables in port operation should include liquid goods and the number of travelers. Liu (1995) used the ownership of port operation, port size, port location and density of capital as dummy variables to evaluate the technical efficiency in England. 5

6 3. DATA SOURCE AND DEFINITION OF VARIABLES In ASEAN and V.I.S.T.A., the container port throughput and equipments data in Lao is not available because it was a land-locked country. Therefore Lao was not included in this research. By the way, the container data in Myanmar in 2007 was also not available, so the authors have deleted this DMU in the analysis. In addition, ASEAN and V.I.S.T.A. are both include Vietnam and Indonesia, therefore the DMUs included in this analysis are Indonesia, Malaysia, Singapore, Thailand, Philippine, Vietnam, Cambodia, Brunei, South Africa and Argentina. It was few to explore about the competitiveness in emerging countries by previous scholars, and study on the efficiency in emerging nations port operation was even less, so this study chosen V.I.S.T.A. and ASEAN as research targets, and analyze the efficiency of the largest port in these countries. From the literature review section, the authors have summarized the possible input variables and output variables related with the ports operation, and presented them in table 2. By using the input-oriented model, it was found that most of the literatures indicated terminal area, the length of terminal, the number of gantry crane, the number of yard crane and the number of straddle carrier could be used as input variables. Therefore we had chosen these variables to be our input variables. Concerning the output variables, the number of container was most frequently reported in previous studies, so it was treated as the output variable in the analysis. Before DEA analysis, the data about input and output variables are required to meet the rule of isotonicity. It means the output should be increased when input had increased. Therefore we examined the relationship between input variable and output variable through Pearson s correlation coefficient which presented in table 1. It was found there was negative relationship between the number of straddle carrier and the number of TEUs. In order to satisfy the rule of isotonicity, the authors have deleted the number of straddle carrier from the input variables. After the variable was deleted, the number of DUMs should be at least twice larger than the sum of the number of input variables and output variables. It is found that the requirement on this rule of thumb is met in this research Table 1 the correlation coefficient analysis between inputs and outputs Input Output The number of container throughput (TEUs) Terminal area (ha) Total length of terminals (m) The number of quayside gantries (number) The number of yard gantries (number) The number of straddle carriers (number)

7 Table 1 Input variables reported in previous ports efficiency studies by DEA technique Scholars Tongzon (2001) Itoh (2002) Barros & Athanassiou (2004) Cullinane et al. (2005) Barros (2006) Cullinane & Wang (2006) Kaisar et al. (2006) Lin & Tseng (2007) Wu & Lin (2008) Yip, Sun & Liu (2010) Input variables Terminal area The number of berth The number of quayside gantries Labor Total length of berth The number of yard gantries The number of straddle carriers The number of deepwater terminal The number of employee Capital Operational cost Fixed capital The waiting time to ship The number of tugs Source: compiled by this research. 7

8 Source: compiled by this research. Table 2 Output variables reported in previous ports efficiency studies by DEA technique` Scholars Output variables Tongzon (2001) Itoh (2002) Barros & Athanassiou (2004) Cullinane et al. (2005) Barros (2006) Cullinane & Wang (2006) Kaisar et al. (2006) Lin & Tseng (2007) Wu & Lin (2008) The number of arrival ships The number of container throughput Liquid bulk Dry bulk Number of ships Passengers Containers with TEU Containers without TEU Sales Ship s working rate The number of cargo throughput The movement about cargo in terminal Yip, Sun & Liu (2010) 8

9 4. METHODOLOGY According to how the DEA model is orientated, DEA model could be classified into input orientated model and output orientated model. Input orientated is associated with operation and management, and output orientated is associated with planning and strategy (Cullinane et al., 2005). As globalization trend and international trade volumes expand rapidly, ports should inspect their capacity frequently to ensure that the service they provided to satisfy customer s demand and maintain their competitiveness. From this viewpoint, this study used DEA-CCR and DEA-BCC model which were under output orientated to evaluate the efficiency of major container ports in ASEAN and V.I.S.T.A. nations. DEA-CCR model was proposed by Charnes et al. (1978), they suppose product process is a constant return to scale, the portfolio of production will increase or decrease proportionally with return to scale which increase or decrease. Put simply, CCR model assumes constant return to scale (CRS) and efficient DMUs must exhibit CRS (Zhu, 2000). CCR model is a combination of a technical efficiency measure and a scale efficiency measure. Banker et al. (1984) have expanded the concept of constant return to scale by proposing the concept of variable return to scale. But DEA-CCR and DEA-BCC couldn t analyze the ranking of container ports, or distinguished their advantage and shortage relatively. Cross-efficiency analysis means it was evaluated by other DMUs input and output weight (Doyle & Green, 1994). DEA-CCR and DEA-BCC will normally define too many efficient ports, so the efficiency of DMUs were evaluated by calculating cross-efficiency matrix to distinguish high efficient DMUs and low efficient DMUs, and defined the DMU which had good performance in overall efficiency. It also could be used to find the relative ranking of each DMU. In addition to the efficiency value, it also illustrated objectivity of evaluating weight. Cross-efficiency is also called peer-assessment. 5. RESEARCH FINDINGS This section introduced DEA-CCR, DEA-BCC and cross-efficiency model to analyze the efficiency of the largest port in each of the ASEAN and the V.I.S.T.A. nations and applied a sensitivity analysis technique to understand the major factors influenced these ports operating efficiency. Finally, the relative location of DMUs in a magic quadrant was depicted through evaluating the absolute efficiency values in the self-assessment (CCR) model and the relative efficiency value was calculated by relative assessment (CCR cross-efficiency) model. 5.1 DEA-CCR, DEA-BCC, Cross-Efficiency under DEA-CCR and DEA-BCC model Table 4 show the port efficiency of these countries in our sample which were analyzed by DEA-CCR, DEA-BCC, DEA-CCR cross-efficiency and DEA-BCC cross-efficiency models in In DEA-CCR model, it was found port of Singapore, port of Ambarli, and port of Durban were efficient ports. The efficiency value of the others were not over 0.8, and the port of Buenos Aires(0.3144)in Argentina and port of Sihanoukville(0.3085)in Cambodia were the less inefficient ports. In DEA-BCC model, in addition to port of Singapore, port of Ambarli in Turkey and port of Durban in the South Africa were 1, the efficiency value of port of Muara in Brunei and port of Klang in Malaysia were also 1 (see table 4), the efficiency value in DEA-BCC is higher than the efficiency value in DEA-CCR in general. It is because the DEA-CCR model assumes the constant return to scale, and this assumption was not existed in the DEA-BCC model. However, the correlation coefficient is as high as between the DEA-CCR model and the DEA-BCC model. 9

10 It also could see there are two types of return to scale in eleven ports in our sample from table 4. The ports revealed constant return to scale included Philippine, Singapore, Thailand, Vietnam, South Africa and Turkey. It implied these ports current production scale is optimal. The ports revealed increasing return to scale included Argentina, Brunei, Cambodia, Indonesia and Malaysia. It implied their current production scale is not the optimal and can be increased to improve their operation efficiency. Another type of return to scale might be decreasing return to scale, but it is not found in this study. Doyle and Green (1994) used DEA-CCR self-assessment model to analyze which DMUs were efficient. However the DMUs in the CCR model did not take peer appraisals into consideration, it evaluates the performance of each DMU by self appraisal. When a DMU was analyzed under cross-efficiency model, its efficiency value is seriously dropped. In other words, if a DUM has the largest difference between self-assessment efficiency and relativeassessment efficiency, it implied this DMU was an outlier. Table 4 is a combination of DEA- CCR and DEA-CCR cross-efficiency statistics which indicate there was no outlier in our ten DMUs. 5.2 Sensitivity analysis This study used sensitivity analysis to examine the level of change of efficiency value by deleting input variables individually in turn. The input variable is more important had the efficiency value largely decreased after the variable is deleted. Model Country Table 4 Compare the efficiency value under 4 DEA models CCR BCC Scale CCR Cross Score Rank Score Rank RTS efficiency efficiency BCC Cross efficiency Outlier 1 Argentina ICR Brunei ICR Cambodia ICR Indonesia ICR Malaysia ICR Philippines CSR Singapore CSR Thailand CSR Vietnam CSR South Africa CSR Turkey CSR Table 5 shows that the number of yard gantries is a important input variable for the majority of these ten ports, these include the most important port in Argentina, Brunei, Cambodia, Indonesia, Malaysia, Philippine, Thailand, Vietnam and South Africa. In the Turkey port, the efficiency value was 1 originally, but if the variable, the number of quayside gantries, was deleted, the original value efficiency value 1.0 will drop to It could infer the number of quayside gantries is an important variable for the Turkish port to maintain its operating efficiency. 1 Outlier = (CCR-CCR Cross Efficiency)/CCR Cross Efficiency 10

11 Table 5 Sensitivity analysis DMU CCR (2005) Terminal area Efficiency value after input deleted Total quay length Number of quayside gantries Number of yard gantries Argentina Brunei Cambodia Indonesia Malaysia Philippines Singapore Thailand Vietnam South Africa Turkey Comprehensive analysis between self-assessment and relative-assessment under DEA-CCR model (Efficiency group in ports) According to self-appraisal (absolute efficiency value) and relative-appraisal (relative efficiency value) under DEA-CCR model, it was found they had positive relationship and correlation coefficient is This study used these efficiency values to depict scatter plot in order to understand relative location of efficiency value of DMUs. This scatter plot can be used to improve the ports efficiency. The self-assessment values have been divided into three tiers, less favorable self-assessment group, favorable self-assessment group and more favorable self-assessment group, along the y axis. Similarly, the relative-assessment value have been divided into three tiers by the efficiency value of 0.6 and 0.8, they were less favorable relative-assessment group, favorable relative-assessment group and more relative-assessment group respectively. Base on the above-mentioned rules to depict each of the ten DMUs efficiency in the figure 1. GroupⅠ- (Self-assessment from CCR efficiency scores =1, relative-assessment from CCR crossefficiency scores > 0.8 ): Ports of the Group Ⅰ are treated as benchmark DMUs mainly. Their efficiency is better than ports in the other two groups. These ports have outperformed the other ports in most of their operational index. It includes the port of Singapore and the port of Durban, these two ports were referenced by the other DMUs by 6 times and 7 times. They should maintain their current advantageous position. GroupⅡ- (Self-assessment from CCR efficiency scores < 1, relative-assessment from CCR crossefficiency scores between 0.6 and 0.8): Ports of this group have an average performance, and they might have an above average performance on some of the efficiency appraisal indices, but they still have not performed well in the other indices which require invest resources to make further improvement. The port of Klang is the only DMU belonged to this group. 11

12 GroupⅢ- (Self-assessment from CCR efficiency scores < 1, relative-assessment from CCR crossefficiency scores < 0.6 ): Ports in the group III didn t have any significant competitive advantage over their peering ports. These ports should manage to understand the sources of their operational inefficiency, and did more effort to improve their competitiveness. The countries of this group include Philippine, Thailand, Cambodia, Brunei, Vietnam, Indonesia and Argentina. Surprisingly, the authors found most of the member nations in the ASEAN and the V.I.S.T.A. were belonged to this group. It implied these economics have not cared too much on the operational efficiency of their ports. Of these seven nations in the group III, Philippine, Vietnam, and Cambodia with their per capita national GDP USD$2011, $1168, and $1035 in 2009 are not belonged to the emerging economics yet. Brunei has the highest nominal GDP USD$26,325 among the seven nations in Thailand, Indonesia, and Argentina have their per capita nominal GDP USD$4,620, $2,858, and $8,663 respectively in As globalization and regionalization tendency prevails, international shipping traffics are increased at a phenomenal speed. How to choose an adequate calling port is a key issue for ocean container carriers. Extant researches on port selection have either sent their questionnaires to freight forwarders or ocean container carriers. From the viewpoint of forwarder, Tongzon (2008) found that ports competitiveness and efficiency come from six key aspects, including high efficiency, good geographic position, low port charge, sufficient infrastructures, broad service scope, and good connectivity with other ports. Among these six key aspects, port efficiency is the most important factors to the freight forwarders. From ocean carriers viewpoint, Chang et al. (2008) indicated six factors are concerned by carriers in selecting their calling port, the port s hinterland cargo volume, port charge, the availability of berth, geographical location, the transit and transshipping cargo quantity, and the connectivity between ports. From both the forwarders and carriers viewpoint, port charge has the highest concern from its users aspect. Some scholars claim that port charge (including terminal handling charge and port disbursement) is the most important factor influencing their port choice decision-making (Lirn, Thanopoulou, Beynon, & Beresford, 2004; Ng, 2006). Arvis et al. (2007) have proposed a Logistics Performance Index and Indicators table (see table 6) which include port charge on a forty feet container movement as a key indice. In this research, the authors find some ports in our analysis, as shown in figure 1, have lowered their port charge to improve their poor logistics performance index resulting from their inadequate operation efficiency, insufficient infrastructure, and inappropriate geographic location. For example, major ports in the following nations have low port charge to attract ocean container ship s calling, Indonesian port charge USD$ 266/244(TEU/FEU), Thailand ports charge $USD 422/422(TEU/FEU), Vietnamese ports charge USD$ 197/294(TEU/FEU), and Cambodian ports charge $USD 355/422(TEU/FEU). In this research, the authors find a special DMU in figure 1, the major port of Turkey is an efficient port in self-assessment round evaluation, but its performance in the relativeassessment round evaluation is not as high as it is in the self-assessment round. This Turkish port was not belonged to any of the three groups in the Figure 1. The authors suspect this Turkish port s input variables were not highly regarded by its peer ports in the relativeassessment round evaluation, thus its CCR cross-efficiency score is comparatively lower than its score in the self-assessment round. Figure 1 comprehensive analysis in DEA-CCR and Cross-Efficiency under Fair peer appraisal. Good peer appraisal. 12

13 Poor self-appraisal Good self-appraisal Fair self-appraisal Poor peer appraisal. Number of references 13

14 Table 6 Logistics performance survey question Rate of physical inspection (percent) Customs clearance (days) Lead time export, median case(days) Lead time import. Best case(days) Lead time import, median case(days) Group Ⅰ Number of border agencies, exports Number of border agencies, import Possibility of a review procedure (percent) Typical charge for a 40-foot export container or a semitrailer (US$) Typical charge for a 40-foot import container or a semitrailer (US$) South Africa Singapore Average Group Ⅱ Malaysia Average Source: World Bank (Arvis et al., 2007) 14

15 Table 6 Logistics performance survey question (Continued) Rate of physical inspection (percent) Customs clearance (days) Lead time export, median case(days) Lead time import. Best case(days) Lead time import, median case(days) Group Ⅲ Number of border agencies, exports Number of border agencies, import Possibility of a review procedure (percent) Typical charge for a 40-foot export container or a semitrailer (US$) Typical charge for a 40-foot import container or a semitrailer (US$) Philippines Thailand Brunei N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A Argentina Vietnam Indonesia Cambodia Average Source: World Bank (Arvis et al., 2007); N/A= data not available 15

16 6. CONCLUSION Firstly, this study has used four models in the DEA technique, including DEA-CCR, DEA-BCC, cross-efficiency in DEA-CCR, and cross-efficiency in DEA-BCC models, to explore the port operation efficiency and analyze their performance according to selfassessment and relative-assessment rules. The authors have used statistics data in Containerisation International Yearbook 2006 to evaluate these ports operation efficiency. Using these four models, this research result has found port of Singapore, port of Ambarli and port of Durban are efficient ports, and these ports could keep their status quo. Secondly, this study has made a comprehensive performance evaluation on these ports by self-assessment and relative-assessment rule, and classified these ports into three groups. Group Ⅰ: Singapore and South Africa are the best ports both in the self-assessment round evaluation and in the relative-assessment round evaluation. They could be used as the benchmark ports. Those inefficient ports in this study could study these benchmark ports practices to develop their own business investment strategy. From the Logistics Performance Index and Its Indicators table (table 6), it is easy to find that the benchmark, port of South Africa and port of Singapore, have outperformed most of the other ports in group Ⅱ and group Ⅲ on the following indices, including physical goods inspection by the customs, customs clearance, lead-time of import and export. It could be found the benchmark port located in group I, port of Singapore and port of South Africa had advantageous geographical position. Singapore is situated in the major shipping route between Europe and Asia, and South Africa is one of important economies in Africa and its port is also located in the major trade route. These findings also evidence that port location is important for a port s overall performances, and it could influence a port s operation efficiency. (Chang et al., 2008) Except the port of Singapore and the major South African port, under their current container throughput volume, other ports in our study should increase their input on the operational equipments, public facilities and changing the size of their quayside terminal area. Without these suggested improvements, the ever-growing import and export containers throughput will generate operational bottlenecks and serious congestion. And according to the table 6 (the Logistics Performance Index and its Indicators), it was found the physical inspection by customs in Philippine was as high as 32%. This ratio is the highest among these emerging economics in this study. Management level of the Philippine port should pay close attention to this phenomena, the improvement on the customs physical inspection rate could largely increase its port efficiency. Previous researches have proposed lower port charge could increase ocean carriers willingness to call the port (Lirn et al., 2004; Ng, 2006), and this argument is supported by this research too. In this research, some of the ports in the emerging economics have insufficient infrastructures and inadequate geographical position, and thus their port operational efficiency is poor. However they did intend to improve their overall performance by lowering port charges on import and export cargoes to attract carriers visit. Ports in the Indonesia, Thailand, and Cambodia are some of the examples. Some researches have found inadequate highway connection and inland transportation infrastructure, poor industrial cluster, and expensive fuel supply could all result in operational inefficiency to container ports (Chandrasekaran & Kumar, 2004). A port s operational efficiency could be greatly increased by improving inefficient inland transport network and by constructing quayside rail transport system (Kaisar et al., 2006). Previous literatures have also revealed the size of a port will influence its operational efficiency. The bigger the ports are, the higher the operational efficiency they have. It is easier for big ports to achieve the advantages resulting from economics of scale and economics of scope (Barros, 2006; Turner, 2004). In the other hand, some previous researches 16

17 have claimed the development of a port does not always necessarily require the port s expansion. According to its location, its characteristics, and its market demands, a port s efficiency can be improved by balancing the resources it inputs and the output it received, and also by an appropriate marketing positioning strategy (Itoh, 2002). Thus, those ports of the group Ⅲ in the Figure 1 can follow this strategy to make their port develop plan and to improve their operational efficiency. ACKNOWLEDGEMENTS Authors thanks for constructive comments from Mr. Lin Chia-Wen during this research. Thanks are also extended to Taiwan s National Science Council for it financially sponsored this research under the NSC H project. REFERENCES 1. Anonymous (2008), JF ASEAN Funds - a global growth engine, accessed f on 30 June Arvis, J.-F., Mustra, M. A., Panzer, J., Ojala, L., & Naula, T. (2007). Connecting to compete trade logistic in the global economy (Publication., from Word Bank: 3. Banker, R. D., Charnes, A., & Cooper, W. W. (1984). Some model for estimating technical and scale inefficiencies in Data Envelopment Analysis. Management Science, 30(9), Barros, C. P. (2006). A benchmark analysis of Italian seaports using data envelopment analysis. Maritime Economics & Logistics, 8, Barros, C. P., & Athanassiou, M. (2004). Efficiency in European seaports with DEA: Evidence from Greece and Portugal. Maritime Economics & Logistics, 6(2), Chang, Y.-T., Lee, S.-Y., & Tongzon, J. L. (2008). Port selection factors by shipping lines: Different perspectives between trunk liners and feeder service providers. Marine Policy, In press, Corrected Proof, Available online 22 July Charnes, A., Cooper, W. W., & Rhodes, E. (1978). Measuring the efficiency of decision making units. European Journal of Operational Research, 2(6), Containerization International Yearbook. ( ). 9. Cullinane, K., & Song, D.-W. (2003). A stochastic frontier model of the productive efficiency of Korean container terminals. Applied Economics, 35(3), Cullinane, K., Song, D.-W., & Wang, T. (2005). The application of mathematical programming approaches to estimating container port production efficiency. Journal of Productivity Analysis, 24(1), Cullinane, K., Wang, T.-F., & Ji, P. (2005). The relationship between privatization and DEA estimates of efficiency in the container port industry. Journal of Economics and Business, 57(5), Cullinane, K. P. B., & Wang, T.-F. (2006). The efficiency of European container ports: A cross-sectional data envelopment analysis. International Journal of Logistics: Research and Applications, 9(1), Dowd, T., & Leschine, T. (1990). Container terminal productivity: A perspective. Maritime Policy and Management, 17(2), Doyle, J., & Green, R. (1994). Efficiency and cross-efficiency in DEA: derivations, meanings and uses. Journal of the operational research society, 45(5),

18 15. González, M. M., & Trujillo, L. (2007). Reform and infrastructure efficiency in Spain's container ports. Transportation Research Part A, In Press, Corrected Proof, Available online 22 October Hsieh J.Y. (2008) The transfer of the BRICK to VISTA. Journal of Excellence, 277, accessed on 30 July Huang S.N. (1999). The application of two stages DEA model on performance evaluation- An study on the environmental protection organizations in Taiwan. Management and System, 6(1), pp Itoh, H. (2002). Efficiency changes at major container ports in Japan: A window application of data envelopment analysis. Review of Urban & Regional Development Studies, 14(2), Kaisar, E. I., Pathomsiri, S., & Haghani, A. (2006). Efficiency measurement of US ports using data envelopment analysis. Paper presented at the National Urban Freight Conference. 20. Kim, M., & Sachish, A. (1986). The structure of production, technical change and productivity in a port. The Journal of Industrial Economics, 35(2), Lin, L. C., & Tseng, C. C. (2007). Operational performance evaluation of major container poerts in the Asia-Pacific region. Maritime policy & Management, 34(6), Lirn, T., Thanopoulou, H., Beynon, M., & Beresford, A. (2004). An application of AHP on transhipment port selection: A global perspective. Maritime Economics & Logistics, 6, Liu, Z. (1995). The comparative performance of public and private enterprises: The case of British ports. Journal of Transport Economics and Policy, 29(3), Martinez-Budria, E., Diaz-Armas, R., Navarro-Ibanez, M., & Ravelo-Mesa, T. (1999). A study of the efficiency of Spanish port authorities using data envelopment analysis. International Journal of Transport Economics XXVI (2), Ng, K. Y. (2006). Assessing the attractiveness of ports in the north European container transhipment market: An agenda for future research in port competition. Maritime Economics & Logistics, 8, Roll, Y., & Hayuth, Y. (1993). Port performance comparison applying data envelopment analysis (DEA). Maritime Policy and Management, 20(2), Tongzon, J. (2001). Efficiency measurement of selected Australian and other international ports using data envelopment analysis. Transportation Research Part A 35(2), Tongzon, J. L. (2008). Port chose and freight forwarders. Transportation Research Part E, In press, Corrected Proof, Available online 22 July Valentine, V. F., & Gray, R. (2001). The measurement of port efficiency using data envelopment analysis. Paper presented at the in Proceedings of the 9th World Conference on Transport Research. 30. Wang Y.J. (2008) Vietnam, Indonesia, South Africa, Turkey, and Argentina the VISTA five, a new global growth engine. Journal of the Future, 261, accessed on 30 July Wu, Y. C., C. W. Lin (2008), National Port Competitiveness: Implications for India, Management Decision, (Accepted). 32. Yip T.L., Sun X.Y., & Liu J.J. (2010) Port Benchmarking: Advantages of Global Terminal Operators. Gwangyang, Korea: The proceedings of the 3 rd International Conference of the Asian Journal of Shipping & Logistics (pp ), Apr , Zhu Joe (2000) Further discussion onlinera production functions and DEA, European Journal of Operational Research, 127(3), pp