Efficiency of Container and Bulk Cargo Ports

she'd be free for lunch from 12:45pm-2:30pm or anytime between 4pm-6pm.
Efficiency of world ports in container and

bulk cargo (oil, coal, ores and grain)
Please cite this paper as:
Merk, O., Dang, T. (2012) Efficiency of world ports in

container and bulk cargo (oil, coal, ores and grain), OECD
Regional Development Working Papers, 2012/09, OECD
Publishing
http://dx.doi.org/10.1787/5k92vgw39zs2-en
OECD Regional Development Working

Papers, 2012/09
Olaf Merk, Thai Thanh Dang
JEL classification: R41, R11, L91
OECD REGIONAL DEVELOPMENT WORKING PAPERS
This series is designed to make selected OECD studies on regional development issues available to a wider
readership. Authorship is usually collective, but principal authors are named. The papers are generally
available in their original language, either English or French, with a summary in the other if provided.
This work is published on the sole responsibility of the author(s). The opinions expressed and arguments
employed herein do not necessarily reflect the official views of the OECD or of the governments of its
member countries.
Comment on the series is welcome, and should be sent to [email protected] or the Public Governance
and Territorial Development Directorate, 2, rue Andr Pascal, 75775 PARIS CEDEX 16, France.
--------------------------------------------------------------------------OECD Regional Development Working Papers are published on

www.oecd.org/gov/regional/workingpapers
--------------------------------------------------------------------------Applications for permission to reproduce or translate all or part of this material should be made to: OECD
Publishing, [email protected] or by fax +33 1 45 24 99 30.
OECD 2012
Photo credits: (from left to right) Hamburg Ports Authority, Port of Caen-Ouistreham, GPMR R.
Hondier
ABSTRACT
Port efficiency is an important indicator of port performance; more efficient ports lower transportation
costs and facilitate imports and exports of a country. Despite the importance of the subject, the exisiting
port efficiency studies have almost exclusively focused on container ports. This Working Paper aims to fill
that gap by calculating efficiency scores of world ports per cargo type (containers, oil, coal, iron ore and
grain). These calculcations have been made using a database constructed for this purpose. Several findings
can be derived from these calculations. Significant improvements can be made when the technical
efficiency of ports is increased. Among the sample, gaps between terminal efficiency mostly reflected gaps
in pure technical efficiency. When comparing the level of efficiency achieved by ports across
commodities, technical gaps were more marked for container and oil terminals. Promoting policies to raise
throughput levels in order to minimise production scale inefficiencies is another important area for
improvement. Production scale inefficiencies arise when throughput levels are below or above optimal
levels given the current capacity of terminal infrastructure. Such inefficiencies were mostly found in a
substantial number of ports handling crude oil and iron ore, suggesting that efficiency is more sensitive and
driven by exogenous factors related to traffic flows. The analysis also shows that the size of ports matters
for port efficiency. The crude oil, iron-ore and grain ports have higher efficiency scores at larger total port
size, suggesting that this size is more efficient because they can drive technological development. Finally,
there are regional patterns emerging across commodities. Terminals in China are among the most efficient
in handling coal bulk and containers with terminals in Southeast Asia. By contrast, the most efficient grain
and iron-ore terminals are located in Latin America, and the most efficient crude-oil transhipment
terminals are mostly found in the Gulf region. Further, Australia is also found to perform well in handling
coal bulk and grains.
JEL classification: R41, R11, L91
Keywords: port efficiency, ports, transportation
FOREWORD
This working paper is one in a series of OECD Working Papers on Regional Development published
by the OECD Public Governance and Territorial Development Directorate. It forms part of the OECD Port
Cities Programme. This paper was written by Olaf Merk, (Administrator, OECD Regional Development
Policy Division) and Thai Thanh Dang (Consultant). The paper benefits from data collection by Xiao
Wang, Mathieu Bordes and Rachel Silberstein. The publication was edited by Caitlin Connelly.
The paper can be downloaded on the OECD website: www.oecd.org/regional/portcities
Further enquiries about this work in this area should be addressed to:
Olaf Merk ([email protected]), OECD Public Governance and Territorial Development Directorate
TABLE OF CONTENTS
1. INTRODUCTION .......................................................................................................................................5
2. THE METHODOLOGICAL APPROACH TO MEASURE PORT EFFICIENCY ...................................6
3. MAIN RESULTS ........................................................................................................................................9
4. CONCLUSIONS .......................................................................................................................................28
BIBLIOGRAPHY .........................................................................................................................................29
ANNEX TABLES .........................................................................................................................................31
NOTES ..........................................................................................................................................................45
1. INTRODUCTION
Port efficiency is an important factor to stimulate port competitiveness and boost regional
development. With growing international sea traffic and changing technology in the maritime transport
industry (containerisation, integrated logistic services, etc.), seaports are coping with mounting pressures to
upgrade and provide cutting-edge technology. They are also being forced to improve port efficiency to
provide comparative advantages that will attract more traffic. Some of the key challenges ports are
surmounting to secure traffic flows and prevent diversion to nearby ports include handling containers and
goods more rapidly, providing more adequate and performing equipment, reducing berth times and delays,
enabling large storage capacity and ensuring multi-modal connections to hinterland. The benefits of port
efficiency extend beyond traffic volume: they have direct and indirect effects on related activities, such as
maritime insurance, finance, and logistics, because of their strategic position in the transport chain. They
create value added and employment, which affect the prospect of regional and urban growth.
Seaport efficiency is often associated with productivity and performance; however, their focus is
narrow, measuring operating technology or total traffic volumes of seaports, which are not the only
indicators. There are additional factors that are associated with the more organisational side of production,
such as how efficiently ports use inputs to produce current output levels and whether the technologies
adopted by ports are the most efficient, that are critical to determining port efficiency.
The indicator of productive port efficiency is thus defined along: i) an efficient production frontier,
which maximises port output for different input levels; ii) a benchmark of best practices, based on ports
located on the efficient production frontier; iii) observable gaps between what ports currently produce and
what they would optimally produce if they were operating efficiently. The efficiency indicator is really a
measure of existing inefficiencies, which is assessed against the most efficient ports and relative
differences in adopted technologies, production scales and input utilisations. This Working Paper should
indicate where potential efficiency gains could be improved, providing insights for guiding development
policy strategies that yield more efficient ports.
One important aspect in measuring port efficiency that is often overlooked in related literature is
related to the multi-activity nature of ports. With the exception of a study by Oliveira and Cariou (2011),
the existing literature tends to measure port efficiency based on volume and technology at container
terminals or the containerised part of ports. Although container activity represents an important segment of
many ports, the measurement of global port efficiency should not be limited to this port use. Other port
activities, which can be very large for some ports, include dry bulk, liquid bulk and general cargo traffic,
and to a lesser extent, passenger traffic. In 2009, average liquid and dry bulk cargo throughput at the top 20
EU ports amounted to 65% of total port traffic in tonnage against 27% for containers, according to
Eurostat-statistics.
This report intends to achieve three main objectives: i) reflect, with the use of a new database, port
multi-activities in assessing port efficiency; ii) estimate port efficiency by activity (e.g. container, oil, coal,
iron ore and grain) and identify the key performance factors, along technology, scale and input efficiencies;
and iii) discuss some general issues on regional and port size patterns across world port efficiency.
2. THE METHODOLOGICAL APPROACH TO MEASURE PORT EFFICIENCY
2.1 Methodology
The methodology used to measure efficiency relies on the data envelopment analysis (DEA)
technique. This empirical methodology derives efficiency scores for each decision-making unit (DMU)
involved in a homogeneous production process such as firms or seaports. An efficient port is defined as
one maximising output level for the same level of inputs across all observed ports (efficient output-oriented
DMU) or minimising quantity of inputs for a given level of output (efficient input-oriented DMU). The
efficient production frontier is delineated by a set of efficient DMUs referred to as the benchmark of most
performing seaports. The potential gains for less efficient ports (e.g. located below the efficient production
frontier) are measured by their distance, both from an output- or input-oriented approach, relative to the
efficiency frontier. This methodology has been widely used in the most recent mainstream literature1
2
(Cheon, et al., 2010; Wu and Goh, 2010; Martinez-Budria, et al., 1999; Wang and Cullinane, 2006; AlEraqui, et al., 2007; Tongzon, 2001).
The DEA approach has advantages as well as limitations. Among its positive characteristics, DEA
does not impose any functional form to the production function or on the shape of returns to scale (i.e. nonparametric), such as when adopting a Cobb Douglas production function. For seaports, in particular, it is
very difficult to guess or impose whether returns to scale should be increasing or decreasing. Dealing with
multiple output processes is another useful property of DEA, especially when addressing port multiactivities and when a certain degree of homogeneity in the production process is observable across ports.
DEA also has some negative characteristics, including its deterministic property, which does not allow
random noises or measurement errors to be isolated from the measure of pure inefficiency3. However, use
of the Bonilla (2000) and Barros (2007) bootstrapping4 technique can help limit this effect.
This sampling technique enables generation of a stochastic distribution and intervals of confidence
around the estimators (Simar and Wildon, 2000). The efficiency estimates derived from using this
technique are often lower compared to DEA estimates derived from a standard sample. In addition,
atypical efficient ports (characterised by low density of observations in the region of the frontier) are
characterised by higher degrees of uncertainty. However, because efficiency is a relative measure,
depending on observable seaports and inputs considered, any omission may affect the results. A sample
excluding potentially efficient seaports or including outliers would respectively shift downward/upward on
the efficient production frontier and affect (upward/downward) the relative efficiency scores. To the same
extent, omitting input factors or including them with non-documented values (zero or not available [n.a.])
may yield higher efficiency scores for ports that are using high quantities of the omitted input factor or
those producing output with no input.
There are three different types of efficiency that can be distinguished: i) overall efficiency, ii)
technical efficiency, and iii) scale efficiency.
i) Overall efficiency. This general indicator, derived from a model assuming constant returns to scale
(CRS), provides a measure of overall port efficiency. This DEA-CCR indicator, developed by Charnes,
Coopers and Rhodes (1978), assumes that all observed production combinations could be scaled up and
down proportionally. Varying production sizes or scales are considered to have no effect on efficiency
6
scoring, which means that small or large ports can equally operate in an efficient way. Efficient ports are
both technically and scale efficient. Conversely, inefficiencies (efficiency gap measured in per cent of most
efficient port scores) reflect both technical and scale inefficiencies.
ii) Technical efficiency. Pure technical efficiency is estimated by relaxing the constraint on scale
efficiency, allowing output to vary unproportionally more or less with a marginal increase in inputs. This
DEA-BCC indicator, developed by Banker, Charnes and Cooper (1984), is derived from a model assuming
varying returns to scale (VRS), and recognises that smaller ports may face disadvantages caused by
production scale effects (Cheon, 2008). By taking into account and neutralising scale inefficiencies,
relative gaps in efficiency between ports would thus only reflect differences in operational inefficiency, socalled pure technical inefficiency.
iii) Scale inefficiencies. Scale inefficiencies arise when the scale of production is inappropriate, being
above or below optimal levels and generating production wastes. Formally, they are identified when a
difference appears between efficiency achieved at technical and overall levels, as measured by the
following ratio (Cooper, et al., 2000; see also Fare, et al., 1994).5
SE=CRS/VRS and where SE<1
In the equation, CRS and VRS are the efficiency estimates derived from respectively assuming
constant and varying returns to scale. When SE<1, ports face scale inefficiency, driving higher overall
inefficiency compared to pure technical inefficiency. By contrast, when SE=1, ports are operating at
efficient scales, producing at the optimal level for which they were designed. However, the appropriate
direction in scale adjustments can be identified only with the nature of returns to scale, that is, increasing
(IRS) or decreasing (DRS). For ports operating at IRS (output rises proportionally more than the increase
in inputs), production level should be expanded. This is usually the case for ports operating below optimal
levels as long as current business traffic, while building up gradually, remains below the optimal capacity
of port infrastructure. By contrast, when ports operate at DRS (output rises proportionally less than the
increase in inputs) they should scale down their production toward lower optimal levels to limit
inefficiencies lead, for example, by bottlenecks. In a long-run perspective, however, the alternative of
raising the optimal level of production through investing in higher port infrastructure capacity should also
be considered.
2.2 Input, output variables and dataset
Defining and identifying appropriate output and input variables for port production function is crucial.
The input/output variables must reflect the main objectives of a port, which in this study is about
maximising cargo throughput and productivity while efficiently using infrastructure and equipment. Along
the economic theory, output as measured by handling cargo throughput (loaded/unloaded) depends to the
same extent on labour and capital inputs. In port literature, labour input is known as the most challenging
issue due to lack of data reliability and comparability. One of the main reasons is that port labour
organisation is particularly complex, consisting of different types of full- and part-time contracts and
contracts partly managed by private, public and port authorities, which make it difficult to collect complete
and consistent data. Proxies are often used along the argument that labour is usually closely and negatively
correlated to handling equipment: equipment is thus considered to be a proxy for labour. As such, for this
study the number of loading/unloading equipment from ship-to-quay and quay-to-shore is collected per
port for container terminals and the different dry and liquid bulk cargo terminals (oil, coal, iron ore and
grain). Capital inputs, on the other hand, are more readily available as long as they concern land and
infrastructure. Such inputs mainly include terminal surface, quay length or storage capacity.
The aim of this study to extend the assessment of port efficiency beyond container terminals and
container ports brings with it major complexities with regard to data collection of port output. Earlier
studies focusing on container ports have benefited from relatively comprehensive existing datasets on
container port output, with output measured in twenty-foot equivalent units (TEUs), being the equivalent to
a small container. This measure is widely accepted and administered, which allows for comprehensive
analysis. Such a comprehensive and comparative dataset does not exist for other port cargo categories.
Most port authorities publish their total annual throughput in metric tonnes, often differentiated by
containerised, bulk and general cargo, but rarely in more specific categories. While this study aims to give
port efficiency scores for bulk categories, it acknowledges the major differences that exist in the equipment
needed for the different bulk categories such as coal, iron ore, grain and oil. Not surprisingly, almost all
large ports dealing with bulk have one or more specialised terminals in these different bulk cargo
categories. This makes it possible to collect input data per port for these cargo categories (e.g. by adding up
the equipment for all grain terminals in that port). However, the corresponding output data (e.g. grain
throughput per port) are in many cases lacking or not in the public domain. Despite considerable efforts to
collect comprehensive port throughput data per cargo category, this proved to be impossible
In order to overcome this complexity, this study uses a new output dataset, based on a volume output
measure: aggregated ship volume in deadweight tonnes (dwt) calling each port. These data can be derived
from existing comprehensive databases of vessel movements, which include detailed information on ship
types (including volume), as well as arrival and departure times at the different ports. This approach
assumes that the volume of a ship calling a port is correlated with the number of metric tonnes loaded or
unloaded from that ship. This assumption will hold especially for cargo categories with point-to-point
deliveries, as in most bulk cargo categories, but probably less so for cargo categories or containerised
cargo with service loops in which several ports are called (as it would be likely that some ports in the
loops, serviced by the same vessel, will load/unload more cargo than others in the same loop). For this
reason, in this study the number of TEUs, where available, is also considered as an output indicator. The
availability of information on different ship types in the database, most of these specialised in carrying one
specific cargo type (e.g. ore carriers, crude oil tankers, etc.), makes it possible to estimate the aggregated
ship volume per port and per cargo category. While total dwt calling the port (output measure) is not
perfectly correlated with actual throughput, it is no more imperfect than throughput as reported in metric
tonnes and TEUs. Both methods risk double counting due to variations in port calculation of throughput.
For example, in instances of transport from an inland to a deep-sea terminal (counted as an incoming and
outgoing container in the river terminal and then incoming and outgoing for the deep sea terminal) one
container could end up being counted four times.
For the purpose of this study, a database was built to analyse port efficiency across worldwide ports at
aggregated and disaggregated activity levels, gathering data for the most recent available year (2011). The
database covers approximately 100 ports, including all major container and dry and liquid bulk ports in a
wide range of ports located in almost all OECD and non-OECD countries. Most of the input data are drawn
from Lloyds Port of the World 2011 Yearbook, whereas the Lloyds Marine Intelligence Units (for May
2011) comprehensive database of vessel movements was used to derive output data. Given limitations in
the data and the DEA methodology, a number of aggregations/approximations were performed in order to
ensure estimate reliability. The input and output variables used to derive efficiency indicators are described
in the following paragraphs on the efficiency per cargo type. The database reflects existing heterogeneity
across equipment and ports into the differences in productivity and thus technology efficiency.
3. MAIN RESULTS
3.1. Container terminals

Container traffic has grown very rapidly over the last decade, increasing by about 9% per year on
average compared to 4% for total seaborne trade (UNCTAD, 2011). While containerised goods broadly
account for one-quarter of total world traffic volume (i.e. manufacturing goods), container traffic remains
one of the most value added and profitable activities. With ports facing growing and lucrative container
traffic, questioning and improving their efficiency in a competitive environment remain a crucial policy
challenge. Based on a worldwide sample of major ports, this section estimates efficiency scores and
ranking for container ports, and explores the ways to improve potential gains.
Sample description
The sample used includes the 63 largest container ports around the world. The regional profile
broadly reflects the worldwide geographic distribution. About half of the container ports are found in Asia
(e.g. 34% in eastern/south-eastern Asia and 19% in western/southern Asia), while the remaining half is
equally split between Europe and America (e.g. respectively 20% each). In terms of traffic volumes, the
sub-sample covers a total of 687 million dwt in 2011 and 287 thousand TEUs in 2009 based on the latest
data available.
Output variables for container ports consider two distinct measures: the volume estimates in
deadweight tonnes (see Section 2.2) and the number of TEUs. The use of multi-output measures is meant
to reconcile both standard analysis based on TEUs (as seen in the literature review) and the methodology
specific to this analysis (inclusion of dwt). While output measures are not strongly correlated (the rank
correlation coefficient is equal to 0.77), the sensitivity analysis shows that the benchmark group remains
broadly the same: among the 15 most efficient ports identified by different output measures, about 10
common ports are found in both groups. Score estimates and the ranking associated to individual ports,
however, differ to some degree (see Table A.1).
Identified input variables are specific to container terminals. Capital inputs are proxied by the
infrastructure of container terminals, such as total quay lengths, terminal surface and the number of reefer
(or plugging) points for refrigerated container ships. Storage capacity, both in TEUs and ha (hectare), has
not been taken into account due to incomplete data. Inputs collected at terminal levels are thus aggregated
at the port level. Labour inputs are proxied by equipment, such as the number of container cranes (e.g. type
of large dockside gantry cranes for loading/unloading intermodal containers from container ships),
including both quay cranes and yard cranes which differ depending on whether the supporting framework
can traverse the length of the quay or yard. The size of container cranes (specific to the size of container
ships such as Panamax, post-Panamax, super-post-Panamax) and handling equipment (e.g. straddle
carriers, sidelifts, reach stackers, or container lorries used to manoeuvre underneath the crane base and
collect the containers) were taken into account.
Table 1. Descriptive statistics of input and output variables of the container ports sample
Container terminal sample
Average
Max
Min
Total sample
Normalised standard deviation
N (non missing)
TEUs 2009 Output May

2011
4,639 10,944,765
25,866 61,351,881
723
34,202
287,601 678,575,427
1
1
62
62
Quay Surface
length terminal
(ha)
4,814
229
19,410
854
540
13
298,476
8,691
1
1
62
38
Reefer
points
1,875
5,444
24
82,501
1
44
Quay
cranes
(no)
45
208
4
2,602
1
58
Yard
cranes
(no)
97
522
1
4,383
1
45
Source: OECD database
Overall efficiency scores

Overall efficiency scores were derived using the DEA methodology (dhat) and plotted next to the
scores derived from the bootstrapping approach (dhat.bc), including intervals of confidence (conf.int) as
shown in Figure 1. The efficiency assessment of one port relative to another is confirmed when both scores
are closed in values and characterised by a small confidence interval. By contrast, a significant gap
between the two different scores is an indicator that the estimates could be sample biased.
To facilitate the interpretation of results, Figure 1 plots the efficiency scores for a sub-sample of
around 40 ports. Those showing the highest scores can be considered the benchmark group or leader
ports, against which the relative scores of the follower ports are assessed. In addition, standardising the
efficiency score plots with a fixed number of ports allows for comparison with efficiency profiles for other
commodities that are derived later on in the study.
Figure 1. Efficiency scores for a sub-sample of container ports (output-dwt, TEUs)
conf.int(crs)
conf.int(crs)
dhat.bc(crs)
dhat(crs)
1.000
0.900
0.900
0.800
0.800
0.700
0.700
0.600
0.600
0.500
0.500
0.400
0.400
0.300
0.300
0.200
0.200
0.100
0.100
0.000
0.000
Jawaharlal Nehru (Ind)
Salalah (Oma)
Khor Fakkan (Uni)
Haifa (Isr)
Kaohsiung (Chi)
Guangzhou (Chi)
Hong Kong (Chi)
Singapore (Sin)
Tianjin (Chi)
Lianyungang (Chi)
Taichung (Chi)
Shenzhen (Chi)
Rotterdam (Net)
Shanghai (Chi)
Tanjung Pelepas (Mal)
Busan (Sou)
Los Angeles (USA)
Jeddah (Sau)
Port Said (Egy)
Jakarta (Ind)
Algeciras (Spa)
Santos (Bra)
Kingston (Jam)
Chennai (Ind)
Marsaxlokk (Mal)
Long Beach (USA)
Dalian (Chi)
Port Klang (Mal)
Bandar Abbas (Ira)
Cape Town (Sou)
Manzanillo (Pan)
Bremerhaven (Ger)
Damietta (Egy)
Tokyo (Jap)
Incheon (Sou)
Antwerp (Bel)
Zeebrugge (Bel)
Xiamen (Chi)
Colombo (Sri)
Gioia Tauro (Ita)
Laem Chabang (Tha)
1.000
Source: Authors own calculations.
10
Note: (dhat) refers to efficiency scores derived using the standard DEA methodology; (dhat.bc) indicates scores derived using the
bootstrapping method and (conf.int) indicates the upper/lower bound values of the interval of confidence; (crs) is the abbreviation of
constant returns to scale, assumption used in both methodologies.
The main features emerging from the efficiency profile of container terminals are (see Annex Table
A.2 for detailed estimates):
Most efficient container ports operate at 70% to 80% of the maximum efficiency level. These
ports are mostly located in Asia, with the exception of Rotterdam. Most efficient ports are operating at high
efficiency values according to both standard and bootstrapped estimates. However, they still operate under
their optimal levels, suggesting that overall efficiency could be improved by 20% to 35% compared to their
current levels. The regional pattern shows that these ports are partly located in western/central Asia and in
China.
Most efficient container ports are not necessarily the largest ports. Among most efficient ports,
are some of the largest global container ports: e.g. Hong Kong, Singapore, Shenzhen and Shanghai
(handling from 20 to 60 million dwt per port per month), but also medium to small size ports (handling 3 to
7 million dwt per port per month). Further, when measuring the rank correlation between efficiency scores
and output, as measured by dwt, the coefficient is 0.27, which indicates that there is not a strong
correlation between terminal/port size and efficiency, at least for containers.
Efficiency potential gains
Figure 2 plots technical and overall efficiency estimates respectively assuming varying and constant
returns to scale. The comparison allows for identification of inefficiency sources: inefficiencies related to
production scale arise when both pure technical estimates (by neutralising production scale inefficiencies)
and overall estimates (without neutralising production scale inefficiencies) differ in values. The gap
indicates the degree by which production scale inefficiencies undermine the level of technological
efficiency. By contrast, inefficiencies related to technology are identifiable by the relative gap in technical
efficiency scores between a given terminal and the benchmark level (the efficiency frontier). The larger the
technical gap, the greater the efforts needed to improve efficiency towards the benchmark level (see Annex
Table A.2 for detailed estimates).
11

Salalah (Oma)
Khor Fakkan (Uni)
Haifa (Isr)
Kaohsiung (Chi)
Guangzhou (Chi)
Hong Kong (Chi)
Singapore (Sin)
Tianjin (Chi)
Lianyungang (Chi)
Taichung (Chi)
Shenzhen (Chi)
Rotterdam (Net)
Shanghai (Chi)
Tanjung Pelepas (Mal)
Busan (Sou)
Los Angeles (USA)
Jeddah (Sau)
Port Said (Egy)
Jakarta (Ind)
Algeciras (Spa)
Santos (Bra)
Kingston (Jam)
Chennai (Ind)
Marsaxlokk (Mal)
Long Beach (USA)
Dalian (Chi)
Port Klang (Mal)
Bandar Abbas (Ira)
Cape Town (Sou)
Manzanillo (Pan)
Bremerhaven (Ger)
Damietta (Egy)
Tokyo (Jap)
Incheon (Sou)
Antwerp (Bel)
Zeebrugge (Bel)
Xiamen (Chi)
Colombo (Sri)
Gioia Tauro (Ita)
Laem Chabang (Tha)
Figure 2. Technical efficiency scores for sub-sample of container ports (output=dwt, TEU)
conf.int(vrs)
conf.int(vrs)
12
dhat.bc(vrs)
dhat.bc(crs)
1.000
1.000
0.900
0.900
0.800
0.800
0.700
0.700
0.600
0.600
0.500
0.500
0.400
0.400
0.300
0.300
0.200
0.200
0.100
0.100
0.000
0.000
Note: (dhat.bc) indicates scores derived using the bootstrapping method and (conf.int) indicates the upper/lower bound values
of the interval of confidence; (crs) and (vrs) respectively refer to constant returns to scale and varying returns to scale,
assumptions used in the methodology.
The technical efficiency profile of container terminals points to several key conclusions:
The technical gap is the major source of inefficiencies across container ports. In most cases,
technical and overall efficiency estimates are almost equal in value indicating that the main source of
inefficiency is closely related to the technical gap. As such, most efficient ports are also technically
efficient. Generally speaking, both follower container ports and most efficient ports have some room to
improve their level of technical efficiency, as most efficient ports still operate between 70% to 80% of the
efficiency frontier.
High technical efficiency is sometimes undermined by production scale inefficiencies. High
technical efficiency scores are mostly concentrated in western/central Asia, Brazil and Jamaica, though not
in the most developed OECD regions. Among these ports, one-third operate at lower (overall) efficiency
levels, reflecting inefficiencies related to inappropriate production levels. For these ports, the adjustments
to limit production scale inefficiencies depend on whether ports are operating at increasing (irs) or
decreasing returns (drs). Table 2 shows the VRS/NIRS ratio, which compares efficiency scores relaxing
the assumption on returns to scale (VRS) and imposing decreasing returns to scale (NIRS). Ports are
operating at decreasing returns when the VRS/NIRS ratio is around the unit value. When the ration differs
from the unit value, ports are operating at increasing returns. As such, Table 2 indicates that the selected
ports are all operating at increasing returns, suggesting that ports are operating under their optimal levels
and that production should be scaled up in order to reduce such inefficiencies.
Table 2. Production returns to scale ratio for selected container ports
Output ranking Port names
54
Damietta (Egy)
53
Chennai (Ind)
46
Incheon (Sou)
61
Cape Town (Sou)
35
Santos (Bra)
43
Kingston (Jam)
vrs
nirs
0.676
0.739
0.630
0.669
0.688
0.677
ratio
0.375
0.472
0.370
0.421
0.491
0.489
13
0.55
0.64
0.59
0.63
0.71
0.72
irs/drs
irs
irs
irs
irs
irs
irs
3.2. Oil terminals

Crude oil, petroleum and liquid gas are among the major commodities handled by seaborne traffic,
amounting to about half of total traffic volume. Fluctuations in world oil supply and demand are important
factors in determining traffic volumes, but they are not the sole factors. Other key challenging factors
include the efficiency of port infrastructure to berth oil tankers, handle liquid oil/gas via pipelines, or to
provide substantial storage capacity and the existence of chemical industries and refineries on port sites.
This section estimates efficiency scores, not yet analysed in literature, for a sample of crude oil ports across
and identifies/discusses the potential sources of improvement for oil port efficiency
Sample description
The sample includes 71 major worldwide ports. The regional pattern reflects a noticeable imbalance
in the distribution of terminals across the world. About two-thirds of the sample oil ports are concentrated
in Asia (with 34% in the East/Southeast and 24% in the western/southern), while the remaining ports are
located in Europe and North America (respectively accounting for 24% and 10% of the total sample).
Table 3 shows input variables specific to the sample oil ports. Capital inputs are proxied by the
capacity of terminal reception of oil tankers, such as quay lengths, maximum vessel capacity, canal
draught/depth and tank storage capacity. Labour input is proxied by the loading capacity of equipment as
measured by their discharge rates (tonne/hour) and pipeline/loading arm capacity (diameter in mm).
Table 3. Descriptive statistics of input/output variables of the crude oil port sample
Oil terminal sample
Average
Max
Min
N (non missing)
Output May Quay length Max vessel

Max
2011
capacity draught/dept
(dwt)
h
2,665,512
33,557,799
2,247
1.98
71
1,833
16,222
100
1.40
52
250,346
750,000
2,000
0.66
47
Source: OECD database.
14
19
50
5
0.44
66
Tank
storage
capacity
(m3)
2,300,030
7,092,000
123,211
1.04
9
Discharge Pipeline/loa
rate (t/h)
ding arm
capacity
(mm)
32,016
9,623
112,000
382
1.13
11
25,245
2,040
0.85
12

Figure 3. Efficiency scores for a sub-sample of crude oil ports/terminals
conf.int(crs)
conf.int(crs)
dhat.bc(crs)
dhat(crs)
1.000
0.900
0.900
0.800
0.800
0.700
0.700
0.600
0.600
0.500
0.500
0.400
0.400
0.300
0.300
0.200
0.200
0.100
0.100
0.000
0.000
Galveston (USA)
Jebel Dhanna Terminal (Uni)
Kharg Is. (Ira)
Juaymah Terminal (Sau)
Ras Laffan (Qat)
Mina al Ahmadi (Kuw)
Fujairah Anch. (Uni)
Ras Tanura (Sau)
Rotterdam (Net)
Primorsk (Rus)
Long Beach (USA)
Yosu (Sou)
Corpus Christi (USA)
Jamnagar Terminal (Ind)
Hong Kong (Chi)
Ningbo (Chi)
Qingdao (Chi)
Pasir Gudang (Mal)
Daesan (Sou)
Novorossiysk/Commercial Sea Port
Zhoushan (Chi)
Ulsan (Sou)
Fos (Fra)
Texas City (USA)
Freeport (Bah)
Cayo Arcas (Mex)
Trieste (Ita)
Algeciras (Spa)
Durban (Sou)
Paradip (Ind)
New Mangalore (Ind)
Kaohsiung (Tai)
Yanbu (Sau)
Bandar Abbas (Ira)
Le Havre (Fra)
Kawasaki (Jap)
Las Palmas (Spa)
Mai-Liao (Tai)
Milford Haven (UK)
Mizushima (Jap)
1.000

constant returns to scale, assumptions used in both methodologies.
Figure 3 plots both the standard DEA and bootstrapped efficiency estimates for crude oil
ports/terminals. The main features include (see Annex Table A.3 for detailed estimates):
Most efficient terminals still have potentials to improve efficiency. Efficient ports are mostly
located in the Gulf region, excluding Galveston (USA) and Rotterdam (NLD), but not all ports in the Gulf
region are operating efficiently (see Mina El Hamadi (KUW), Fujhaira (UAE) and Ras Tamura (SAU)
ports). On average, most efficient terminals achieve around 60% to 70% of the efficiency frontier,
suggesting that 30 % gains in production could be achieved given their existing inputs/infrastructure.
Oil terminal size matters. Most efficient terminals also tend to rank among the top ten terminals in
terms of volume. Rank correlation between port size/volume and port efficiency scores shows a strong and
positive correlation (coefficient of 0.95), suggesting that efficiency is strongly and significantly associated
to oil traffic volumes.
15
Sources of efficiency/inefficiency
Figure 4. Technical efficiency scores for a sub-sample of crude oil ports/terminals
conf.int(vrs)
conf.int(vrs)
dhat.bc(vrs)
dhat.bc(crs)
1.000
0.900
0.900
0.800
0.800
0.700
0.700
0.600
0.600
0.500
0.500
0.400
0.400
0.300
0.300
0.200
0.200
0.100
0.100
0.000
0.000
Galveston (USA)
Jebel Dhanna Terminal (Uni)
Kharg Is. (Ira)
Juaymah Terminal (Sau)
Ras Laffan (Qat)
Mina al Ahmadi (Kuw)
Fujairah Anch. (Uni)
Ras Tanura (Sau)
Rotterdam (Net)
Primorsk (Rus)
Long Beach (USA)
Yosu (Sou)
Corpus Christi (USA)
Jamnagar Terminal (Ind)
Hong Kong (Chi)
Ningbo (Chi)
Qingdao (Chi)
Pasir Gudang (Mal)
Daesan (Sou)
Novorossiysk/Commercial Sea Port (Rus)
Zhoushan (Chi)
Ulsan (Sou)
Fos (Fra)
Texas City (USA)
Freeport (Bah)
Cayo Arcas (Mex)
Trieste (Ita)
Algeciras (Spa)
Durban (Sou)
Paradip (Ind)
New Mangalore (Ind)
Kaohsiung (Tai)
Yanbu (Sau)
Bandar Abbas (Ira)
Le Havre (Fra)
Kawasaki (Jap)
Las Palmas (Spa)
Mai-Liao (Tai)
Milford Haven (UK)
Mizushima (Jap)
1.000

Note: (dhat.bc) indicates scores derived using the bootstrapping method and (conf.int) indicates the upper/lower bound values of the
interval of confidence; (crs) and (vrs) respectively refer to constant returns to scale and varying returns to scale, assumption used in
the methodology.
Marked gaps between technical and overall efficiency estimates as shown in Figure 4 indicate
important sources of inefficiencies related to production scales ranging from point loses of 10% to 40%.
Figure 4 highlights several key findings:
Technical efficiency is the key driver of overall efficiency for most oil ports/terminals. Most
technically efficient ports are also most overall efficient ones. As such, the major source of potential gains
for these ports is to catch-up the technological ladder towards the efficiency frontier. To the same extent, a
similar assessment would also apply to all follower ports/terminals with closed estimates (between
technical and overall scores).
Inefficiencies are also partly driven by production scales. Some of the crude oil ports/terminals are
particularly marked by large gaps between pure technical and overall efficiency scores. These gaps mainly
reflect efficiency losses due to production scales, pointing out that the production level is below/above
optimal values given the existing infrastructure. Production scale inefficiencies are usually found to
negatively affect technical efficiency scores by 40% to 50% points, especially for small (Bangkok [THA],
16
Djakarta [IDN] and London [GBR] and Salalah [OMA] Savannah [USA]) and medium size terminals (Las
Palmas [ESP], Jawaharlal Nehru, New Mangalore [IND] and Texas City [USA]) (see Annex tables A.3).
Increasing oil production volumes towards optimal levels would improve efficiency. The
comparison between efficiency estimates derived by imposing or not decreasing returns indicates that most
crude oil ports/terminals facing production scale inefficiencies are operating at increasing returns to scale.
This suggests that terminal efficiency could be improved by adjusting oil production towards upward
optimum levels.
Table 4. Production returns to scale for selected crude oil ports
vrs
nirs
ratio
24
Texas City (USA)
0.768
0.292
0.38
43
0.630
0.157
0.25
37
New Mangalore (Ind)
0.605
0.154
0.25
41
Las Palmas (Spa)
0.727
0.129
0.18
58
Savannah (USA)
0.730
0.026
0.04
60
Bangkok (Tha)
0.636
0.016
0.03
63
Jakarta (Ind)
0.632
0.014
0.02
66
London (UK)
0.642
0.011
0.02
65
Salalah (Oma)
0.638
0.006
0.01
17
irs/drs
irs
irs
irs
irs
irs
irs
irs
irs
irs
3.3. Coal bulk terminals

Dry bulk goods in seaborne traffic amount to one-quarter of total volume of minerals (coal, iron ore,
etc.) and vegetables (grain, wood, edible foods, etc.). Dry bulk maritime traffic, despite being relatively
less valuable and profitable than crude oil and container traffic, also faces lower entry costs driven by
cutting edge technology used to operate container and liquid bulk terminals. As such, diversification
strategies and fierce competition are likely to be more intense across worldwide ports handing dry bulk
cargoes. The following three sections provide efficiency analysis (e.g. score estimates, ranking and source
of improvement) for ports/terminals handling coal, iron ore and grain bulk cargoes.
Sample description
The coal bulk port sample includes 34 of the largest ports across the world. The regional distribution
of the sample is broadly well balanced across Asia, America and Oceania, while Europe tends to be underrepresented. Capital inputs used by dry bulk terminals dedicated to handling coal are proxied by specific
terminal quay length and their storage capacity. Labour input is approximated by the capacity of terminal
handling equipment ranging from 1 to 39 thousand tonnes per hour over the sample.
Table 5. Descriptive statistics of coal bulk terminals/ports sample
Coal bulk terminal sample
Average
Max
Min
N (non missing)
Output May 2011
2,178,910
7,787,066
41,688
1.00
34
18
Quay length Storage capacity Loading/unloa

(m)
(tonnes)
ding (total
capacity per
h)
1,020
4,215
235
0.94
33
2,648,195
10,425,000
350,000
0.95
28
10,863
39,000
1,000
0.97
27

Figure 5. Efficiency scores for coal bulk terminals
conf.int(crs)
conf.int(crs)
dhat.bc(crs)
dhat(crs)
1.000
0.900
0.900
0.800
0.800
0.700
0.700
0.600
0.600
0.500
0.500
0.400
0.400
0.300
0.300
0.200
0.200
0.100
0.100
0.000
0.000
Hay Point (Aus)

Velsen/Ijmuiden (Net)
Banjarmarsin (Ind)
Shanghai (Chi)
Qinhuangdao (Chi)
Newcastle (Aus)
Rizhao (Chi)
Guangzhou (Chi)
Gladstone (Aus)
Puerto Bolivar (Col)
Oita (Jap)
Sepetiba (Bra)
Taranto (Ita)
Richards Bay (Sou)
Paradip (Ind)
Vancouver (Can)
Alexandria (Egy)
Nagoya (Jap)
Port Kembla (Aus)
Abbot Point (Aus)
Amsterdam (Net)
Tanjung Bara (Ind)
Balikpapan (Ind)
Mobile (USA)
Baltimore (USA)
Long Beach (USA)
Tarragona (Spa)
Hampton Roads/Norfolk (USA)
La Spezia (Ita)
Swinoujscie (Pol)
Brisbane (Aus)
Prince Rupert (Can)
Gdansk (Pol)
Tallinn (Est)
1.000

The main findings shown in Figure 5, which plots both efficiency scores based on standard and
bootstrapped DEA, include:
A clear group of efficient terminals is emerging from the sample, mostly located in Australia
and China. According to both estimates (standard and bootstrapped), about one-third of the coal bulk ports
figures among the benchmark of efficient terminals. Most of these terminals are located in Australia and
China, excluding those located in the Netherlands, India and Colombia (see Annex Table A.4 for detailed
estimates). Operating at 65% to 75% of the efficiency frontier, there is also room to improve operational
efficiency
Follower terminals are significantly lagging behind. Indeed, follower ports/terminals drop
down to less than 40% efficiency, and the least performing terminals fall below 10% efficiency. Further,
many of the least performing ports are located in developed countries in Europe and the USA.
19

Figure 6. Technical efficiency scores for coal bulk terminals
conf.int(vrs)
conf.int(vrs)
dhat.bc(vrs)
dhat.bc(crs)
Tallinn (Est)
Gdansk (Pol)
Brisbane (Aus)
Prince Rupert (Can)
La Spezia (Ita)
Swinoujscie (Pol)
Tarragona (Spa)
Hampton Roads/Norfolk (USA)
Long Beach (USA)
Mobile (USA)
Baltimore (USA)
Balikpapan (Ind)
Amsterdam (Net)
0.000
Tanjung Bara (Ind)
0.000
Abbot Point (Aus)
0.100
Nagoya (Jap)
0.100
Port Kembla (Aus)
0.200
Alexandria (Egy)
0.200
Paradip (Ind)
0.300
Vancouver (Can)
0.300
Taranto (Ita)
0.400
Richards Bay (Sou)
0.400
Oita (Jap)
0.500
Sepetiba (Bra)
0.500
Gladstone (Aus)
0.600
Puerto Bolivar (Col)
0.600
Rizhao (Chi)
0.700
Guangzhou (Chi)
0.700
Newcastle (Aus)
0.800
Qinhuangdao (Chi)
0.800
Shanghai (Chi)
0.900
Banjarmarsin (Ind)
0.900
Hay Point (Aus)
1.000
1.000

the methodology.
Figure 6 highlights the importance of efficiency gains from improved technology (e.g. equipment) and
infrastructure.
The main source of efficiency gains is based on improving the technological ladder of the
infrastructure and equipment. For almost all coal bulk ports, overall and pure technical estimates are
found to be very closed in values, indicating the absence of inefficiencies driven by non-optimal
production levels. As such, the main driving factor of overall efficiency is related to gaps in pure technical
efficiency. Improving the technological ladder appears as a key lever, including for terminals located in
most developed countries. This assessment excludes a few terminals (Alexandria [EGY], Toranto [ITA]
and Paradip [IND]) where efficiency gains rely on the adjustment of production scales towards optimal
levels.
20
3.4. Iron bulk terminals

Sample description
The sample of iron bulk terminals includes only 15 ports/terminals around the world, reflecting the
availability of the data. The regional distribution is relatively well balanced, though Africa is underrepresented.
Input variables specific to bulk terminal infrastructure dedicated to iron ore are proxied by quay
lengths and the maximum depth of canals for bulks carriers. For a limited number of terminals, inputs also
include storage capacity. Labour input is proxied by the equipment capacity in loading/unloading minerals
ranging from 1.5 to 60 thousand tonnes per hour.
Table 6. Descriptive statistics of grain ports/terminals sample
Iron bulk terminal sample
Average
Max
Min
N (non missing)
Output May
2011
771,650
3,039,118
100,912
1.11
15
21
Quay Maximum Loading/u

length depth (m) nloading
capacity
(t/h)
981
2,825
168
0.70
15
18
24
10
0.28
15
Storage
capacity
(tonnes)
15,820 2,061,250
60,000 2,975,000
1,500 1,150,000
1.10
0.39
13
4

Figure 7. Efficiency scores for iron bulk terminals
conf.int(crs)
conf.int(crs)
dhat.bc(crs)
dhat(crs)
0.300
0.200
0.200
0.100
0.100
0.000
0.000
Port Hedland (Aus)
0.300
Kashima (Jap)
0.400
Guangzhou (Chi)
0.400
Hamburg (Ger)
0.500
0.500
Seven Islands (Can)
0.600
Visakhapatnam (Ind)
0.600
Dampier (Aus)
0.700
Saldanha Bay (Sou)
0.700
Shanghai (Chi)
0.800
Tubarao (Bra)
0.800
Gladstone (Aus)
0.900
Sepetiba (Bra)
0.900
Ponta da Madeira (Bra)
1.000
Port Walcott (Aus)
1.000

Figure 7 represents efficiency scores based on standard and bootstrapped DEA. Its main findings
highlight a strong relationship between terminal size and efficiency of iron ore terminals.
Most efficient iron ore ports/terminals are mostly large. Achieving 70% to 80% of the frontier of
efficiency, the scores suggest potential room for improvement. These ports principally appear in the top
five terminals in terms of iron ore transhipment volume. The rank correlation between port volume and
efficiency score is around 0.72, which confirms a strong link between the size/volume and efficiency gains
for iron ore ports. However, some caution should be used when considering the results given the relatively
small size of the sample (see Annex Table A.5 for detailed estimates).
22

Figure 8. Technical efficiency scores for iron bulk terminals
conf.int(vrs)
conf.int(vrs)
dhat.bc(vrs)
dhat.bc(crs)
0.300
0.200
0.200
0.100
0.100
0.000
0.000
Port Hedland (Aus)
0.300
Kashima (Jap)
0.400
Guangzhou (Chi)
0.400
Hamburg (Ger)
0.500
0.500
Seven Islands (Can)
0.600
Visakhapatnam (Ind)
0.600
Dampier (Aus)
0.700
Saldanha Bay (Sou)
0.700
Shanghai (Chi)
0.800
Tubarao (Bra)
0.800
Gladstone (Aus)
0.900
Sepetiba (Bra)
0.900
Ponta da Madeira (Bra)
1.000
Port Walcott (Aus)
1.000

the methodology.
Technical efficiency scores show how production scale inefficiencies are an important factor in the
creation of efficiency gaps across iron ore ports/terminals (see Figure 8).
Production scale inefficiencies are major source of efficiency gaps across iron ore
ports/terminals. With the exception of a few cases, iron ore ports are mostly operating at high levels of
technical efficiency, achieving more than 70% of the frontier. The interesting finding is that the overall
efficiency is proportionally decreasing while inefficiencies in production scale are increasing (as identified
by the gaps between both technical and overall estimates). This indicates that, on the basis of the sample,
the major driver for improving efficiency of iron bulk ports/terminals is to reduce production scale
inefficiencies rather than improving technology. As seen in Table 7, showing state of production returns to
scale, adjustments for most of these ports must focus on increasing output volumes as all concerned
terminals are found to operate at increasing returns to scale.
23
Table 7. Production returns to scale for iron ore commodities for selected terminals/ports

12
6
13
14
10
15
Shanghai (Chi)
Saldanha Bay (Sou)
Visakhapatnam (Ind)
Seven Islands (Can)
Hamburg (Ger)
Guangzhou (Chi)
vrs
0.731943
0.732501
0.734986
0.735622
0.739479
0.732503
24
nirs
0.506149
0.47488
0.293871
0.239893
0.198004
0.163193
ratio
drs/irs
0.692
0.648
0.400
0.326
0.268
0.223
irs
irs
irs
irs
irs
irs
3.5. Grain terminals

Sample representativity and input proxies
The data sample covers 41 grain ports/terminals worldwide. The sample is equally distributed across
the main regions, such as Asia, North and South America, followed by Oceania, Europe and to a smaller
extent, Africa. However, the sample is marked by a greater volatility in output around the mean (as
indicated by the normalised standard deviation compared to other commodities) suggesting that the sample
may reflect very large imbalances in size across grain ports/terminals.
Input variables collected are specific to grain terminals. Capital inputs are proxied by quay lengths,
and grain storage capacity, and labour input is proxied by loading grain equipment as measured by the
loading capacity ranging from 400 tonnes to 20 000 tonnes loaded per hour.
Table 8. Descriptive statistics of grain ports/terminals sample
Grain terminal sample
Average
Max
Min
N (non missing)
Output May 2011 Quay length

(m)
769,881
3,450,208
4,942
1.26
41
25
656
3,484
100
1.04
39
Storage
Loading
capacity capacity (per h)
(Tonnes)
413,097
2,470,000
27,945
1.38
33
4,963
20,000
400
1.05
36

Figure 9. Efficiency scores for grain terminals
upper value
lower value
efficiency (boost)
efficiency (std)
1.000
0.900
0.900
0.800
0.800
0.700
0.700
0.600
0.600
0.500
0.500
0.400
0.400
0.300
0.300
0.200
0.200
0.100
0.100
0.000
0.000
Tubarao (Bra)
Chiba (Jap)
Duluth (USA)
Bahia Blanca (Arg)
Busan (Sou)
San Lorenz-San Martin (Arg)
Vancouver (Can)
Thunder Bay (Can)
Rouen (Fra)
Rio Grande (Bra)
Mobile (USA)
Kashima (Jap)
Incheon (Sou)
Shanghai (Chi)
Mizushima (Jap)
Taichung (Tai)
Rosario (Arg)
Houston (USA)
Rotterdam (Net)
Kuwait (Kuw)
Santos (Bra)
Montreal (Can)
Freemantle (Aus)
Hachinohe (Jap)
Geraldton (Aus)
Port Said (Egy)
Beaumont (USA)
Lake Charles (USA)
Ghent (Bel)
Baie Comeau (Can)
Tianjin (Chi)
Port Kembla (Aus)
Newcastle (Aus)
Haifa (Isr)
Buenos Aires (Arg)
East London (Sou)
Portland (Aus)
Brisbane (Aus)
Valencia (Spa)
Southampton (UK)
Trieste (Ita)
1.000

There are two main findings from Figure 9 (see Annex Table A.6 for detailed estimates):
Port size matters. Most efficient terminals are among the top ten largest grain ports/terminals.
There is a clear emerging group of efficient terminals, which according to both standard and bootstrapped
estimates are operating at relatively high standard values. These terminals are also figure among the top ten
largest grain ports/terminals loading more than 1.5 million tonnes of grains per year. However, the
bootstrapped estimates indicate that these terminals operate at more than 55% of the efficiency frontier,
suggesting that the room to increase their efficiency is even larger compared to other commodities. These
terminals are mostly located in Latin America with two terminals in Argentina (Bahia Bianca and San
Lorenz San Martin), in Brazil (Tubarao), Japan (Chiba) and Korea (Busan) but also Duluth in the US.
Terminals in developed countries are poorly performing. Efficiency scores (according to standard
and bootstrapped estimates) rapidly drop down to 30% after the benchmark group and down below 10%
after the first 20 terminals. These relative poorly scores suggest that substantial room exists to improve
efficiency. Surprisingly, least performing terminals tend to be found in developed OECD countries such as
Italy, UK, Spain and Australia.
26
Sources of efficiency/inefficiency
Figure 10. Technical efficiency scores for grain terminals
upper value
lower value
efficiency (boost vrs)
efficiency (boost crs)

1.000
0.900
0.900
0.800
0.800
0.700
0.700
0.600
0.600
0.500
0.500
0.400
0.400
0.300
0.300
0.200
0.200
0.100
0.100
0.000
0.000
Tubarao (Bra)
Chiba (Jap)
Duluth (USA)
Bahia Blanca (Arg)
Busan (Sou)
San Lorenz-San Martin (Arg)
Vancouver (Can)
Thunder Bay (Can)
Rouen (Fra)
Rio Grande (Bra)
Mobile (USA)
Kashima (Jap)
Incheon (Sou)
Shanghai (Chi)
Mizushima (Jap)
Taichung (Tai)
Rosario (Arg)
Houston (USA)
Rotterdam (Net)
Kuwait (Kuw)
Santos (Bra)
Montreal (Can)
Freemantle (Aus)
Hachinohe (Jap)
Geraldton (Aus)
Port Said (Egy)
Beaumont (USA)
Lake Charles (USA)
Ghent (Bel)
Baie Comeau (Can)
Tianjin (Chi)
Port Kembla (Aus)
Newcastle (Aus)
Haifa (Isr)
Buenos Aires (Arg)
East London (Sou)
Portland (Aus)
Brisbane (Aus)
Valencia (Spa)
Southampton (UK)
Trieste (Ita)
1.000
Source: Authorsown calculations.

Note: (dhat.bc) indicates scores derived using the bootstrapping method and (conf.int) indicates the upper/lower bound values
of the interval of confidence; (crs) and (vrs) respectively refer to constant returns to scale and varying returns to scale, assumption
used in the methodology.
The main findings from Figure 10 are:

Improving technical efficiency is the major driver to increase overall efficiency. In most cases,
overall and technical estimates of efficiency have closed values, indicating that overall efficiency reflects
the level of pure technical efficiency. The technical gaps are surprisingly marked across grain
ports/terminals. The correlation between the two efficiency scores is equal to 0.97, which confirms that
upgrading technical efficiency towards the efficiency frontier is a key lever to increase grain port/terminal
efficiency. As a result, a large majority of grain ports/terminals are operating at optimal levels.
27
4. CONCLUSIONS
Port efficiency is a key driver of port competitiveness and can play an important role in boosting
regional development. The analysis in this report provides several insights on where and how to gain port
efficiency.
Technical efficiency is the most important factor to improving port efficiency. Significant
improvements can be made when the technical efficiency of ports is increased. Among the sample, gaps
between terminal efficiency mostly reflected gaps in pure technical efficiency. Most of the performing
ports still have some room to improve pure technical efficiency as long as they are already achieving
between 65% to 75% of the efficiency frontier, whereas follower ports need to catch up the technological
ladder in order to see improvements. When comparing the level of efficiency achieved by ports across
commodities, technical gaps were more marked for container and oil terminals. Most of the performing
container and oil terminals use technology more efficiently compared to iron ore, coal and grain terminals.
This probably reflects the nature of container and crude oil terminals, which when compared to other
commodity terminals are generally more technology embedded and likely to face greater pressures to
provide cutting-edge technology.
Promoting policies to raise throughput levels in order to minimise production scale
inefficiencies is another important area for improvement. Production scale inefficiencies arise when
throughput levels are below or above optimal levels given the current capacity of terminal infrastructure.
Such inefficiencies were mostly found in a substantial number of ports handling crude oil and iron ore,
suggesting that efficiency is more sensitive and driven by exogenous factors related to traffic flows. By
contrast, for containers, grain and coal bulks, production scale inefficiencies were more focused on
individual/specific ports. The handling of these cargo categories were mostly found to operate at increasing
return to scales indicating that throughput levels have to be increased towards higher optimal levels.
However, ports with recent infrastructure investments also face production scale inefficiencies over a
period until current business reach their optimal levels.
The analysis shows that the size of ports matters for port efficiency. The crude oil, iron-ore and
grain ports have higher efficiency scores at larger total terminal size, suggesting that this size is more
efficient because they can drive technological development. However, larger-scale ports are more likely to
operate at decreasing return to scale (e.g. above certain optimal levels) and face overheating inefficiencies.
Policies aimed at relieving traffic congestions in the short run, or increasing terminal infrastructure for
large-scale ports in the long run, would reinforce port efficiency for these specific commodities.
Regional patterns are also seen in port efficiency. There are regional patterns emerging across
commodities. For example, in general, terminals in China are among the most efficient in handling coal
bulk and containers with terminals in Southeast Asia. By contrast, the most efficient grain and iron-ore
terminals are located in Latin America, and the most efficient crude-oil transhipment terminals are mostly
found in the Gulf region. Further, Australia is also found to perform well in handling coal bulk and grains.
28
BIBLIOGRAPHY
Al-Eraqui, A.S., et al. (2007), Evaluating the Location Efficiency of Arabian and African Seaports Using
Data Envelopment Analysis (DEA), Woking Papers Series, WP 019/2007/DE/UECE, School of
Economics and Management, Technical University of Lisbon, Lisbon.
Barros, C.P. and S. Managi (2008), Productivity Drivers in Japanese Seaports, Working Papers Series,
WP 15/2008/DE/UECE, School of Economics and Management, Technical University of Lisbon,
Lisbon.
Bloningen, B.A. and W.W. Wilson (2006), Port Efficiency and Trade Flows, IWR Report 06-NETS-SR-11, Navigation Economics Technologies, www.corpsnets.us.
Banker, R.D., A. Charnes and W. Cooper (1984), Some Models for Estimating Technical and Scale
Inefficiencies in Data Envelopment Analysis, Management Science, Vol. 30, No. 9, pp. 1078-1092.
Charnes, A., W. Cooper and E. Rhodes (1978), Measuring Efficiency of Decision-Making Units,
European Journal of Operational Research, Vol. 2, North-Holland Publishing Company, pp.429444.
Cheon, SH., D.E. Dowall and D.-W. Song (2010), Evaluating Impacts of Institutional Reforms on Port
Efficiency Changes: Ownership, Corporate Structure, and Total Factor Productivity Changes of
World Container Ports, Transportation Research Part E: Logistics and Transportation Review,
Vol. 46, No. 4, pp. 546-561.
Cheon, SH. (2008), Productive Efficiency of World Container Ports: A Global Perspective, Journal of
the Transportation Research Board, No. 2062, Transportation Research Board of the National
Academies, Washington, D.C., pp. 10-18.
Cooper, W.W., L.M. Seiford and K. Tone (2000), Data Envelopment Analysis: A Comprehensive Text with
Models, Applications, References and DEA Solver Software, Kluwer Academic Publisher, Boston.
Fare, R., S. Grosskopf and C.A.K. Lovell (1994), Production Frontiers. Cambridge University Press,
Cambridge.
Gonzales, M.M., and L. Trujillo (2005), Reforms and Infrastructure Efficiency in Spains Container
Ports, World Bank Policy Research, Working Paper 3515, World Bank,
http://econ.worldbank.org.
Gonzales, M.M., and L. Trujillo (2006), Efficiency Measurement in Port Industry: A Survey of Empirical
Evidence, CCRP Working Paper No. 8, City University, London.
Herrera, S. and G. Pang (2008), Efficiency of Infrastructure: The Case of Container Ports, Revista
Economia, Vol. 9, No. 1, pp. 165-194.
Martinez-Budria, E., et al. (1999). A Study of the Efficiency of Spanish Port Authorities Using Data
Envelopment Analysis, International Journal of Transport Economics, Vol. 26, No. 2, pp. 237-253.
Munisamy, S. and G. Singh (2011), Benchmarking the Efficiency of Asian Container Ports, African
Journal of Business Management, Vol. 5, No. 4, pp. 1397-1407.
29
Notteboom, T., C. Coeck, J. Van Den Broeck (2000), Measuring and Explaining the Relative Efficiency
of Container Terminals by Means of Bayesian Stochastic Frontier Models, International Journal of
Maritime Economics, Vol. 2, No. 2, pp. 83-106.
Oliveira, G.F. de, Cariou, P. (2011), A DEA study of the efficiency of 122 iron ore and coal ports and of
15/17 countries in 2005, Maritime Policy & Management, Vol.38, No.1, pp. 727-743
Pallis, A., K. Thomas, P.W. De Langen and T. Notteboom (2001), Port Economics, Policy and
Management: Content Classification and Survey, Transport Reviews, Vol. 31, No. 4, pp. 445-471.
Park, R.-K. and P. De (2004), An Alternative Approach to Efficiency Measurement Seaports, Maritime
Economics and Logistics, Vol. 6, No. 1, Kluwer Academic Publishers, Boston, pp.53-69.
Simar, L. and P.W. Wilson (2000), Statistical Inference in Nonparametric Frontier Models: The State of
the Art, Journal of Productivity Analysis, Vol. 13, No. 1, pp.49-78.
Tongzon, J. (2001), Efficiency Measurement of Selected Australian and Other International Ports Using
Data Envelopment Analysis, Transportation Research Part A: Policy and Practice, Vol. 35, No. 2,
pp. 107-122.
Tongzon, J. (1995), Determinants of Port Performance Efficiency, Transportation Research Part A:
Policy and Practice, Vol. 29, No. 3, pp. 245-252.
Wang, T.-F. and K. Cullinane (2006), The Efficiency of European Container Terminals and Implications
for Supply Chain Management, Maritime Economics & Logistics, Vol. 8, pp. 82-99.
Wilson, P.W. (1993), Detecting Outliers in Deterministic Nonparametric Frontier Models with Multiple
Outputs, Journal of Business & Economic Statistics, Vol. 11, No. 3, pp. 319-323.
Wilson, P.W. (2008), FEAR: A Software Package for Frontier Efficiency Analysis with R, SocioEconomic Planning Sciences, Vol. 42, No. 4, pp. 247-254.
Wu, Y.C.J. and M. Goh (2010), Container Port Efficiency in Emerging and More Advanced Markets,
Transportation Research Part E: Logistics and Transportation Review, Vol. 46, No. 6, pp.10301042.
30
ANNEX TABLES
Table A.1: Sensitivity analysis for DEA scores1 container terminals
Correlations
A/B
DMU
(A)
all ports
0.640
Model A (dwt,
teu)
CRS
CRS std
boot
Score Score2
Country
Port
India
Jawaharlal
Nehru
0.952
0.836
Oman
United Arab
Emirates
Salalah
1.000
0.790
Khor Fakkan
0.971
0.765
57
Israel
Haifa
0.955
0.752
50
Chinese Taipei
Kaohsiung
0.986
0.749
21
24
31
Rank
CRS
Boots
(A)
DMU
(B)
44
0.557
Model B (dwt)
CRS
CRS
Std
Boot
Score Score2
0.559
Rank
CRS
Boots
(B)
Comm
on
ports
in the
efficie
ncy
bench
mark
0.302
0.228
35
1.000
0.701
1.000
0.826
1.000
0.718
28
1.000
0.679
11
79
106
China
Guangzhou
1.000
0.723
20
0.532
0.426
26
China
Hong Kong
1.000
0.715
21
1.000
0.637
14
Singapore
Singapore
0.915
0.696
91
0.727
0.538
20
China
Tianjin
1.000
0.680
26
1.000
0.765
29
China
Lianyungang
1.000
0.676
10
22
0.181
0.143
44
55
Chinese Taipei
Taichung
0.832
0.675
11
30
1.000
0.757
China
Shenzhen
1.000
0.670
12
24
1.000
0.642
12
Netherland
Rotterdam
1.000
0.667
13
77
1.000
0.641
13
China
1.000
0.666
14
23
1.000
0.608
17
34
Malaysia
Shanghai
Tanjung
Pelepas
0.777
0.638
15
0.381
0.277
South Korea
Busan
0.781
0.600
16
93
1.000
0.781
13
USA
Los Angeles
0.715
0.593
17
116
0.772
0.548
18
27
Saudi Arabia
Jeddah
0.710
0.560
18
89
1.000
0.689
26
Egypt
Port Said
0.630
0.547
19
36
0.279
0.219
36
23
Indonesia
Jakarta
0.591
0.510
20
46
0.092
0.064
53
28
Spain
Algeciras
0.630
0.501
21
95
0.712
0.547
19
35
Brazil
Santos
0.644
0.495
22
12
1.000
0.726
43
Jamaica
Kingston
0.613
0.493
23
58
0.583
0.491
22
53
India
Chennai
0.626
0.473
24
43
0.207
0.153
42
34
Malta
Marsaxlokk
0.528
0.461
25
74
0.153
0.110
49
14
73
15
USA
Long Beach
0.552
0.460
26
115
0.260
0.211
37
19
China
Dalian
0.551
0.450
27
19
0.003
0.002
62
11
Malaysia
Port Klang
0.530
0.433
28
72
0.660
0.502
21
37
Iran
Bandar Abbas
0.559
0.424
29
47
0.575
0.449
24
31
61
South Africa
Cape Town
0.556
0.423
30
92
0.590
0.429
25
49
Panama
Manzanillo
0.442
0.385
31
82
0.003
0.002
61
17
Germany
Bremerhaven
0.500
0.379
32
40
0.951
0.761
54
Egypt
Damietta
0.433
0.377
33
35
0.138
0.096
51
22
Japan
Tokyo
0.477
0.376
34
65
0.525
0.391
28
46
South Korea
Incheon
0.457
0.374
35
94
0.089
0.064
54
10
Belgium
Antwerp
0.471
0.352
36
0.883
0.617
16
33
Belgium
Zeebrugge
0.407
0.351
37
0.163
0.130
47
16
China
Xiamen
0.456
0.345
38
27
0.649
0.464
23
25
Sri Lanka
Colombo
0.385
0.329
39
100
0.172
0.135
46
30
Italy
Gioia Tauro
Laem
Chabang
0.371
0.312
40
52
0.219
0.166
40
20
Thailand
0.353
0.297
0.252
0.199
39
42
81
0.255
0.203
38
43
98
0.151
0.107
50
0.243
44
113
0.064
0.048
57
0.315
0.243
45
63
0.773
0.619
15
0.274
0.226
46
0.462
0.396
27
40
Panama
Balboa
0.338
0.293
58
Spain
Las Palmas
0.341
0.291
42
USA
Houston
0.289
38
Japan
Nagoya
59
Australia
Brisbane
41
102
41
Spain
Barcelona
0.256
0.225
47
96
0.079
0.060
55
47
USA
Tacoma
0.256
0.211
48
125
0.077
0.058
56
18
USA
New York
0.261
0.209
49
120
0.207
0.159
41
12
Germany
Hamburg
0.268
0.205
50
41
0.512
0.365
30
36
France
Le Havre
0.257
0.192
51
39
0.852
0.684
10
44
USA
Seattle
0.225
0.186
52
124
0.159
0.122
48
39
USA
Oakland
0.229
0.184
53
121
0.420
0.335
33
50
Pakistan
Karachi
0.212
0.174
54
80
0.188
0.140
45
45
Chinese Taipei
Keelung
0.224
0.170
55
29
0.445
0.339
32
52
Saudi Arabia
Dammam
0.204
0.166
56
88
0.044
0.033
59
60
Lebanon
Beirut
0.209
0.158
57
69
0.437
0.349
31
56
USA
Charleston
0.188
0.147
58
110
0.454
0.370
29
32
USA
Savannah
0.191
0.146
59
123
0.202
0.145
43
48
Italy
Genoa
0.153
0.128
60
51
0.122
0.094
52
62
Japan
Hakata
0.146
0.112
61
60
0.063
0.044
58
51
Canada
Montreal
0.137
0.107
62
14
0.036
0.026
60
Notes:
1
2
Authors' calculation
Ports ranked by bootstrapped DEA scores
32
Table A.2: DEA scores1 under varying assumptions on returns to scale for container terminals
DEA Model specification

Outputs: y1=containers (DWT) , y2=containers (TEUs),
Inputs: x1=quay length, x2=surface terminal, x3=refeer points, x4=quay cranes, x5=yard cranes
Constant returns to scale (CRS)

Stand
ard
DEA
Bootstrapped DEA
Varying returns to scale (VRS)

Stand
ard
DEA
Bootstrapped DEA
Non increasing returns to scale

(NIRS)
Stand
ard
DEA
Bootstrapped DEA
DMU
Country
Port
Score
Score
2
Upper
value
Lower
value
Score
Score
Upper
value
Lower
value
Score
Score
Upper
value
Lower
value
Singapor
e
Singapore
0.915
0.696
0.867
0.405
1.000
0.742
0.962
0.397
1.000
0.735
0.952
0.394
China
Shanghai
1.000
0.666
0.970
0.153
1.000
0.673
0.964
0.162
1.000
0.660
0.960
0.149
China
Hong Kong
1.000
0.715
0.952
0.340
1.000
0.718
0.965
0.330
1.000
0.711
0.950
0.336
China
South
Korea
Shenzhen
1.000
0.670
0.969
0.198
1.000
0.677
0.969
0.198
1.000
0.676
0.957
0.205
Busan
0.781
0.600
0.752
0.347
0.783
0.596
0.759
0.345
0.781
0.588
0.749
0.326
China
Netherlan
d
Guangzhou
1.000
0.723
0.965
0.334
1.000
0.724
0.963
0.334
1.000
0.718
0.972
0.331
Rotterdam
1.000
0.667
0.954
0.191
1.000
0.674
0.964
0.159
1.000
0.670
0.955
0.190
China
Chinese
Taipei
Tianjin
1.000
0.680
0.960
0.210
1.000
0.676
0.956
0.219
1.000
0.671
0.962
0.220
Kaohsiung
0.986
0.749
0.966
0.347
0.986
0.752
0.959
0.338
0.986
0.741
0.970
0.346
10
Belgium
Antwerp
0.471
0.352
0.449
0.168
0.471
0.356
0.452
0.174
0.471
0.353
0.454
0.174
11
Malaysia
Port Klang
0.530
0.433
0.513
0.315
0.532
0.429
0.512
0.306
0.532
0.425
0.510
0.302
12
Germany
Hamburg
0.268
0.205
0.258
0.124
0.349
0.266
0.340
0.149
0.349
0.261
0.334
0.141
13
USA
0.715
0.593
0.688
0.409
0.715
0.593
0.689
0.390
0.715
0.589
0.687
0.396
5
6
7
8
9
Malaysia
Los Angeles
Tanjung
Pelepas
0.777
0.638
0.747
0.455
0.777
0.639
0.746
0.462
0.777
0.632
0.753
0.437
15
USA
Long Beach
0.552
0.460
0.536
0.317
0.552
0.459
0.538
0.304
0.552
0.457
0.534
0.310
16
China
0.456
0.345
0.443
0.168
0.570
0.435
0.558
0.228
0.456
0.343
0.441
0.172
14
Germany
Xiamen
Bremerhave
n
0.500
0.379
0.482
0.189
0.500
0.379
0.485
0.182
0.500
0.374
0.479
0.179
18
USA
New York
0.261
0.209
0.251
0.148
0.276
0.218
0.268
0.144
0.276
0.214
0.264
0.140
19
China
0.551
0.450
0.529
0.313
0.555
0.455
0.533
0.324
0.551
0.449
0.532
0.311
0.353
0.297
0.338
0.232
0.366
0.299
0.352
0.222
0.366
0.295
0.350
0.213
India
Dalian
Laem
Chabang
Jawaharlal
Nehru
0.952
0.836
0.925
0.673
0.968
0.818
0.938
0.631
0.952
0.829
0.928
0.652
22
Japan
Tokyo
0.477
0.376
0.463
0.229
0.480
0.376
0.466
0.228
0.477
0.373
0.460
0.227
23
Indonesia
Jakarta
0.591
0.510
0.570
0.418
0.597
0.500
0.577
0.397
0.591
0.501
0.565
0.399
24
Oman
Salalah
1.000
0.790
0.965
0.529
1.000
0.757
0.971
0.434
1.000
0.785
0.945
0.517
25
Sri Lanka
Colombo
0.385
0.329
0.371
0.267
0.387
0.328
0.373
0.261
0.387
0.328
0.377
0.262
26
Egypt
Saudi
Arabia
Port Said
0.630
0.547
0.608
0.451
0.647
0.551
0.626
0.446
0.630
0.540
0.603
0.427
Jeddah
0.710
0.560
0.677
0.370
0.743
0.554
0.709
0.286
0.710
0.556
0.684
0.353
Spain
Algeciras
Lianyungan
g
0.630
0.501
0.609
0.312
0.652
0.512
0.627
0.300
0.630
0.492
0.611
0.303
China
1.000
0.676
0.953
0.280
1.000
0.675
0.960
0.162
1.000
0.677
0.954
0.271
Italy
United
Arab
Gioia Tauro
Khor
Fakkan
0.371
0.312
0.359
0.246
0.374
0.307
0.360
0.238
0.371
0.307
0.356
0.234
0.971
0.765
0.929
0.452
1.000
0.673
0.952
0.161
0.971
0.756
0.930
0.443
17
20
21
27
28
29
30
31
Thailand
33
Emirates
32
USA
Savannah
0.191
0.146
0.185
0.082
0.191
0.147
0.185
0.083
0.191
0.144
0.184
0.079
33
Belgium
Zeebrugge
0.407
0.351
0.389
0.285
0.408
0.344
0.390
0.281
0.407
0.344
0.393
0.277
34
Malta
Marsaxlokk
0.528
0.461
0.509
0.372
0.548
0.463
0.529
0.360
0.528
0.458
0.510
0.354
35
Brazil
Santos
0.644
0.495
0.632
0.271
1.000
0.688
0.956
0.237
0.644
0.491
0.622
0.262
36
France
0.257
0.192
0.250
0.083
0.257
0.190
0.252
0.071
0.257
0.189
0.251
0.082
Iran
Le Havre
Bandar
Abbas
0.559
0.424
0.535
0.259
0.559
0.425
0.540
0.232
0.559
0.423
0.535
0.247
38
Japan
Nagoya
0.315
0.243
0.305
0.132
0.321
0.244
0.311
0.126
0.315
0.241
0.308
0.133
39
USA
Oakland
0.229
0.184
0.221
0.124
0.229
0.184
0.222
0.121
0.229
0.183
0.220
0.126
40
Panama
Balboa
0.338
0.293
0.330
0.233
0.381
0.321
0.364
0.262
0.338
0.290
0.328
0.233
41
Spain
Barcelona
0.256
0.225
0.250
0.182
0.273
0.236
0.265
0.192
0.256
0.223
0.249
0.182
42
USA
Houston
0.289
0.243
0.279
0.199
0.303
0.252
0.292
0.206
0.289
0.241
0.276
0.199
43
Jamaica
Kingston
0.613
0.493
0.586
0.344
1.000
0.677
0.957
0.154
0.613
0.489
0.594
0.329
44
USA
Chinese
Taipei
South
Korea
Seattle
0.225
0.186
0.217
0.131
0.228
0.189
0.220
0.131
0.225
0.186
0.217
0.132
Keelung
0.224
0.170
0.216
0.086
0.224
0.169
0.216
0.086
0.224
0.167
0.216
0.087
Incheon
0.457
0.374
0.441
0.269
0.812
0.630
0.779
0.402
0.457
0.370
0.442
0.268
47
USA
Tacoma
0.256
0.211
0.250
0.146
0.259
0.211
0.252
0.141
0.256
0.211
0.249
0.149
48
Italy
Genoa
0.153
0.128
0.147
0.099
0.153
0.127
0.149
0.095
0.153
0.125
0.145
0.094
49
Panama
Manzanillo
0.442
0.385
0.424
0.315
0.468
0.369
0.454
0.244
0.442
0.381
0.423
0.299
50
Pakistan
Karachi
0.212
0.174
0.205
0.120
0.213
0.174
0.205
0.119
0.212
0.172
0.206
0.113
51
Canada
Saudi
Arabia
Montreal
0.137
0.107
0.132
0.070
0.137
0.107
0.132
0.066
0.137
0.106
0.131
0.067
Dammam
0.204
0.166
0.196
0.121
0.221
0.183
0.216
0.139
0.221
0.182
0.214
0.138
53
India
Chennai
0.626
0.473
0.614
0.266
1.000
0.739
0.966
0.288
0.626
0.472
0.602
0.273
54
Egypt
Chinese
Taipei
Damietta
0.433
0.377
0.421
0.309
1.000
0.676
0.969
0.168
0.433
0.375
0.420
0.307
Taichung
0.832
0.675
0.795
0.492
0.863
0.643
0.827
0.280
0.832
0.669
0.804
0.470
56
USA
Charleston
0.188
0.147
0.182
0.085
0.189
0.146
0.178
0.087
0.188
0.144
0.181
0.084
57
Israel
Haifa
0.955
0.752
0.909
0.502
1.000
0.670
0.958
0.161
0.955
0.746
0.923
0.497
58
Spain
Las Palmas
0.341
0.291
0.330
0.232
0.422
0.332
0.406
0.211
0.341
0.290
0.332
0.226
59
Australia
Brisbane
0.274
0.226
0.265
0.154
0.334
0.254
0.327
0.123
0.274
0.223
0.265
0.148
60
Lebanon
South
Africa
Beirut
0.209
0.158
0.202
0.079
0.212
0.155
0.202
0.065
0.209
0.157
0.199
0.081
Cape Town
0.556
0.423
0.536
0.268
1.000
0.669
0.956
0.156
0.556
0.421
0.528
0.262
Japan
Hakata
0.146
0.112
0.142
0.066
0.177
0.140
0.170
0.085
0.177
0.140
0.170
0.096
37
45
46
52
55
61
62
Note
s:
1
Author's calculation
Ports ranked by Bootstrapped DEA scores
34
Table A.2 (cont'd): DEA scores1 under varying assumptions on returns to scale for container terminals

Output: y1=containers (DWT)
Inputs: x1=quay length, x2=surface terminal, x3=refeer points, x4=quay cranes, x5=yard cranes

Standa
rd DEA
DMU

Stand
ard
DEA
Bootstrapped DEA
Bootstrapped DEA

(NIRS)
Stand
ard
DEA
Bootstrapped DEA
Country
Port
Score
Score
2
Upper
value
Lower
value
Score
Score
Upper
value
Lower
value
Score
Score
Upper
value
Lowe
r
value
Argentina
Buenos
Aires
0.398
0.283
0.384
0.079
0.422
0.293
0.407
0.071
0.398
0.287
0.386
0.077
Australia
Brisbane
0.462
0.396
0.448
0.307
0.600
0.493
0.583
0.333
0.462
0.391
0.438
0.305
Australia
Melbourne
0.152
0.119
0.139
0.082
0.156
0.118
0.142
0.077
0.156
0.122
0.144
0.084
Australia
Newcastle
0.011
0.008
0.010
0.006
0.091
0.059
0.086
0.005
0.011
0.008
0.010
0.005
Australia
Port Kembla
0.002
0.002
0.002
0.001
0.002
0.001
0.002
0.001
0.002
0.002
0.002
0.001
Australia
Banglades
h
Sydney
0.005
0.004
0.005
0.002
0.005
0.004
0.005
0.002
0.005
0.004
0.005
0.002
Chittagong
0.049
0.038
0.045
0.022
0.049
0.037
0.045
0.022
0.049
0.038
0.046
0.023
Belgium
Antwerp
0.883
0.617
0.840
0.234
0.918
0.624
0.846
0.224
0.883
0.613
0.836
0.234
Belgium
Zeebrugge
0.163
0.130
0.154
0.099
0.166
0.127
0.153
0.091
0.163
0.128
0.154
0.093
10
Brazil
Itajai
0.587
0.447
0.555
0.289
0.587
0.410
0.548
0.127
0.587
0.444
0.551
0.286
11
Brazil
Navegantes
0.115
0.082
0.109
0.039
0.115
0.081
0.111
0.038
0.115
0.083
0.111
0.039
12
Brazil
Santos
1.000
0.726
0.942
0.390
1.000
0.642
0.929
0.174
1.000
0.728
0.942
0.385
13
Brazil
Sepetiba
0.146
0.120
0.143
0.074
0.147
0.119
0.145
0.070
0.146
0.121
0.143
0.077
14
Canada
Montreal
0.036
0.026
0.034
0.012
0.036
0.026
0.033
0.012
0.036
0.026
0.034
0.012
15
Canada
Vancouver
1.000
0.714
0.923
0.402
1.000
0.599
0.926
0.003
1.000
0.704
0.936
0.364
16
Chile
San Antonio
0.048
0.038
0.044
0.028
0.062
0.049
0.058
0.034
0.062
0.050
0.058
0.036
17
Chile
Valparaiso
0.527
0.429
0.501
0.286
0.527
0.400
0.498
0.230
0.527
0.429
0.498
0.294
18
China
Beilun
0.727
0.539
0.684
0.288
0.763
0.532
0.709
0.242
0.763
0.550
0.730
0.276
19
China
Dalian
0.003
0.002
0.002
0.001
0.003
0.002
0.002
0.001
0.003
0.002
0.002
0.001
20
China
Guangzhou
0.532
0.426
0.499
0.281
0.532
0.419
0.500
0.283
0.532
0.429
0.496
0.303
21
China
1.000
0.637
0.949
0.173
1.000
0.623
0.920
0.138
1.000
0.635
0.944
0.177
China
Hong Kong
Lianyungan
g
0.181
0.143
0.178
0.075
0.183
0.141
0.179
0.073
0.181
0.144
0.177
0.074
23
China
Shanghai
1.000
0.608
0.935
0.075
1.000
0.599
0.924
0.038
1.000
0.605
0.919
0.073
24
China
Shenzhen
1.000
0.642
0.926
0.180
1.000
0.634
0.936
0.155
1.000
0.643
0.913
0.187
25
China
Taicang
0.042
0.034
0.040
0.023
0.042
0.032
0.039
0.021
0.042
0.034
0.040
0.023
China
Tianjin
1.000
0.765
0.944
0.395
1.000
0.748
0.941
0.377
1.000
0.768
0.951
0.394
China
Chinese
Taipei
Chinese
Taipei
Chinese
Taipei
Xiamen
0.649
0.464
0.594
0.221
0.700
0.480
0.650
0.194
0.649
0.458
0.609
0.215
Kaohsiung
1.000
0.679
0.909
0.292
1.000
0.673
0.932
0.283
1.000
0.688
0.954
0.295
Keelung
0.445
0.339
0.421
0.204
0.445
0.332
0.415
0.195
0.445
0.343
0.422
0.209
Taichung
1.000
0.757
0.936
0.510
1.000
0.675
0.929
0.235
1.000
0.762
0.950
0.502
Colombia
0.778
0.604
0.717
0.428
0.778
0.551
0.715
0.264
0.778
0.605
0.728
0.411
Costa Rica
Cartagena
Puerto
Limon
0.203
0.150
0.190
0.092
0.235
0.162
0.216
0.076
0.203
0.150
0.193
0.092
Ecuador
Guayaquil
0.175
0.138
0.166
0.090
0.177
0.133
0.163
0.082
0.175
0.137
0.166
0.092
22
26
27
28
29
30
31
32
33
35
34
Egypt
Alexandria
0.052
0.036
0.051
0.008
0.052
0.035
0.050
0.008
0.052
0.036
0.050
0.009
35
Egypt
Damietta
0.138
0.096
0.132
0.033
0.186
0.127
0.175
0.038
0.138
0.096
0.131
0.035
36
Egypt
Port Said
0.279
0.219
0.259
0.146
0.285
0.217
0.262
0.139
0.285
0.224
0.268
0.151
37
Estonia
Tallinn
0.030
0.022
0.028
0.012
0.051
0.035
0.046
0.015
0.030
0.022
0.028
0.011
38
France
Dunkirk
0.183
0.148
0.173
0.095
0.183
0.140
0.171
0.082
0.183
0.149
0.173
0.098
39
France
0.852
0.684
0.821
0.402
0.852
0.671
0.823
0.377
0.852
0.689
0.825
0.423
Germany
Le Havre
Bremerhave
n
0.951
0.761
0.906
0.473
0.951
0.743
0.893
0.442
0.951
0.764
0.910
0.494
41
Germany
Hamburg
0.512
0.365
0.472
0.159
0.524
0.362
0.481
0.142
0.524
0.371
0.485
0.147
42
Greece
Piraeus
0.169
0.127
0.159
0.069
0.169
0.124
0.159
0.066
0.169
0.127
0.162
0.071
43
India
Chennai
Jawaharlal
Nehru
Kolkata
(Calcutta)
0.207
0.153
0.194
0.073
0.294
0.210
0.273
0.076
0.207
0.153
0.194
0.076
0.302
0.228
0.282
0.123
0.308
0.230
0.283
0.122
0.302
0.231
0.284
0.131
0.149
0.119
0.138
0.090
0.434
0.333
0.403
0.185
0.149
0.117
0.137
0.084
0.092
0.064
0.088
0.026
0.184
0.139
0.168
0.083
0.184
0.144
0.171
0.089
0.575
0.449
0.535
0.310
0.575
0.417
0.529
0.226
0.575
0.449
0.539
0.312
40
44
45
46
India
India
Iran
Jakarta
Bandar
Abbas
Bandar
Imam
Khomeini
0.121
0.096
0.111
0.069
1.000
0.600
0.921
0.003
0.121
0.093
0.114
0.063
49
Israel
Ashdod
0.664
0.545
0.633
0.374
0.753
0.558
0.727
0.284
0.664
0.541
0.635
0.373
50
Israel
Haifa
1.000
0.718
0.925
0.453
1.000
0.595
0.921
0.017
1.000
0.718
0.946
0.442
51
Italy
Genoa
0.122
0.094
0.112
0.064
0.148
0.113
0.138
0.075
0.148
0.115
0.140
0.081
52
Italy
Gioia Tauro
0.219
0.166
0.206
0.110
0.264
0.199
0.245
0.128
0.264
0.205
0.250
0.139
53
Italy
La Spezia
0.610
0.481
0.573
0.321
0.660
0.488
0.610
0.264
0.610
0.483
0.575
0.315
54
Italy
Leghorn
0.280
0.219
0.262
0.139
0.280
0.210
0.264
0.114
0.280
0.222
0.265
0.136
55
Italy
Taranto
0.152
0.119
0.141
0.081
0.152
0.116
0.136
0.074
0.152
0.120
0.145
0.082
56
Italy
Ivory
Coast
Trieste
0.041
0.029
0.039
0.012
0.046
0.031
0.043
0.012
0.046
0.032
0.043
0.014
Abidjan
0.583
0.441
0.551
0.285
0.583
0.389
0.544
0.091
0.583
0.437
0.552
0.272
58
Jamaica
Kingston
0.583
0.491
0.553
0.387
0.852
0.678
0.806
0.445
0.583
0.481
0.541
0.375
59
Japan
Chiba
0.031
0.022
0.029
0.009
0.041
0.029
0.039
0.009
0.031
0.023
0.029
0.009
60
Japan
Hakata
0.063
0.044
0.061
0.014
0.065
0.044
0.061
0.013
0.063
0.044
0.061
0.015
61
Japan
Kitakyushu
0.081
0.064
0.078
0.043
0.256
0.166
0.243
0.024
0.081
0.061
0.077
0.038
62
Japan
Kobe
0.280
0.195
0.262
0.081
0.287
0.191
0.267
0.064
0.280
0.194
0.264
0.079
63
Japan
Nagoya
0.773
0.619
0.744
0.412
0.794
0.609
0.744
0.382
0.773
0.611
0.721
0.420
64
Japan
Osaka
0.482
0.335
0.461
0.084
0.491
0.333
0.469
0.077
0.482
0.339
0.467
0.082
65
Japan
Tokyo
0.525
0.391
0.502
0.228
0.525
0.380
0.482
0.202
0.525
0.391
0.500
0.225
66
Japan
Yokohama
1.000
0.682
0.915
0.161
1.000
0.665
0.914
0.149
1.000
0.689
0.940
0.158
67
Kenya
Mombasa
0.082
0.061
0.079
0.029
0.082
0.060
0.078
0.027
0.082
0.062
0.079
0.029
68
Kuwait
Kuwait
0.006
0.005
0.006
0.002
0.006
0.004
0.006
0.002
0.006
0.005
0.006
0.002
69
Lebanon
0.437
0.349
0.418
0.217
0.479
0.368
0.448
0.203
0.437
0.349
0.416
0.215
Malaysia
Beirut
Pasir
Gudang
0.047
0.036
0.045
0.021
0.047
0.035
0.043
0.020
0.047
0.036
0.044
0.021
71
Malaysia
Penang
0.282
0.212
0.270
0.132
0.328
0.217
0.307
0.058
0.282
0.212
0.263
0.131
72
Malaysia
0.660
0.502
0.614
0.336
0.896
0.683
0.809
0.447
0.896
0.703
0.858
0.501
Malaysia
Port Klang
Tanjung
Pelepas
0.381
0.277
0.356
0.128
0.381
0.274
0.358
0.131
0.381
0.280
0.359
0.141
74
Malta
Marsaxlokk
0.153
0.110
0.145
0.046
0.153
0.107
0.144
0.045
0.153
0.110
0.145
0.049
75
Mexico
Manzanillo
0.517
0.372
0.488
0.159
0.573
0.398
0.531
0.154
0.517
0.370
0.487
0.159
76
Morocco
Tangier
0.269
0.190
0.253
0.077
0.271
0.187
0.254
0.066
0.269
0.191
0.256
0.080
77
Netherland
Netherland
s
Rotterdam
1.000
0.641
0.939
0.171
1.000
0.611
0.900
0.094
1.000
0.630
0.923
0.158
Amsterdam
0.000
0.000
0.000
0.000
0.005
0.003
0.004
0.001
0.000
0.000
0.000
0.000
79
Oman
Salalah
1.000
0.701
0.951
0.379
1.000
0.661
0.929
0.283
1.000
0.697
0.947
0.382
80
Pakistan
Karachi
0.188
0.140
0.177
0.084
0.191
0.141
0.179
0.085
0.188
0.142
0.176
0.085
81
Panama
Balboa
0.255
0.203
0.237
0.141
0.261
0.195
0.239
0.117
0.261
0.201
0.242
0.131
82
Panama
Manzanillo
0.003
0.002
0.003
0.001
0.003
0.002
0.003
0.001
0.003
0.002
0.003
0.001
47
Indonesia
Iran
48
57
70
73
78
36
83
Peru
Callao
0.198
0.158
0.190
0.103
0.198
0.154
0.187
0.100
0.198
0.158
0.188
0.108
84
Philippines
Manila
0.001
0.001
0.001
0.000
0.001
0.001
0.001
0.000
0.001
0.001
0.001
0.000
85
Romania
Constantza
0.161
0.130
0.155
0.087
0.168
0.129
0.156
0.077
0.161
0.130
0.154
0.087
86
Russia
Novorossisk
St.
Petersburg
0.038
0.029
0.036
0.015
0.038
0.028
0.035
0.015
0.038
0.029
0.037
0.017
0.021
0.017
0.019
0.012
0.031
0.023
0.029
0.015
0.031
0.024
0.029
0.016
Dammam
0.044
0.033
0.041
0.018
0.050
0.036
0.045
0.019
0.050
0.037
0.047
0.020
Jeddah
1.000
0.689
0.930
0.388
1.000
0.657
0.911
0.316
1.000
0.689
0.939
0.397
87
Russia
Saudi
Arabia
Saudi
Arabia
Saudi
Arabia
88
89
90
Yanbu
0.003
0.002
0.003
0.001
0.003
0.002
0.003
0.001
0.003
0.002
0.003
0.001
Singapore
South
Africa
South
Korea
South
Korea
Singapore
0.727
0.538
0.681
0.296
0.926
0.662
0.858
0.355
0.926
0.679
0.881
0.397
Cape Town
0.590
0.429
0.563
0.174
0.787
0.547
0.746
0.142
0.590
0.429
0.558
0.166
Busan
1.000
0.781
0.916
0.517
1.000
0.732
0.927
0.418
1.000
0.755
0.916
0.467
Incheon
0.089
0.064
0.083
0.034
0.092
0.064
0.085
0.030
0.089
0.064
0.083
0.034
95
Spain
Algeciras
0.712
0.547
0.667
0.330
0.712
0.529
0.664
0.286
0.712
0.553
0.684
0.317
96
Spain
Barcelona
0.079
0.060
0.074
0.034
0.115
0.085
0.110
0.047
0.115
0.087
0.111
0.053
97
Spain
Bilbao
0.040
0.028
0.038
0.009
0.050
0.033
0.046
0.008
0.040
0.027
0.038
0.008
98
Spain
Las Palmas
0.151
0.107
0.147
0.040
0.151
0.103
0.143
0.031
0.151
0.106
0.146
0.039
99
Spain
Tarragona
0.090
0.073
0.086
0.043
0.096
0.077
0.093
0.043
0.090
0.073
0.086
0.043
100
Sri Lanka
Colombo
0.172
0.135
0.159
0.092
0.198
0.150
0.182
0.094
0.198
0.155
0.187
0.100
101
Sweden
0.010
0.008
0.010
0.006
0.012
0.010
0.012
0.006
0.010
0.008
0.010
0.006
91
92
93
94
Thailand
Stockholm
Laem
Chabang
0.252
0.199
0.239
0.142
0.315
0.243
0.291
0.168
0.315
0.249
0.297
0.182
103
Turkey
Ambarli
0.483
0.373
0.467
0.195
0.486
0.367
0.471
0.187
0.483
0.375
0.470
0.195
104
Turkey
United
Arab
Emirates
United
Arab
Emirates
United
Kingdom
United
Kingdom
Mersin
0.197
0.156
0.193
0.083
0.199
0.155
0.196
0.081
0.197
0.157
0.193
0.082
Jebel Ali
0.365
0.289
0.352
0.205
0.461
0.357
0.425
0.247
0.461
0.366
0.431
0.266
Khor
Fakkan
1.000
0.826
0.955
0.593
1.000
0.694
0.919
0.330
1.000
0.819
0.946
0.586
Felixstowe
Southampto
n
1.000
0.738
0.935
0.450
1.000
0.707
0.913
0.390
1.000
0.742
0.944
0.465
0.100
0.072
0.096
0.031
0.104
0.072
0.098
0.028
0.100
0.071
0.096
0.031
Uruguay
Montevideo
0.249
0.189
0.234
0.113
0.249
0.184
0.234
0.105
0.249
0.191
0.238
0.118
USA
0.454
0.370
0.426
0.266
0.455
0.359
0.432
0.249
0.454
0.371
0.427
0.271
USA
Charleston
Hampton
Roads
0.869
0.741
0.847
0.518
0.949
0.790
0.927
0.509
0.869
0.742
0.845
0.533
112
USA
Honolulu
0.356
0.278
0.336
0.187
1.000
0.628
0.921
0.079
0.356
0.268
0.334
0.168
113
USA
Houston
0.064
0.048
0.060
0.030
0.071
0.053
0.065
0.032
0.071
0.054
0.066
0.034
114
USA
Jacksonville
0.239
0.203
0.231
0.141
0.259
0.214
0.253
0.137
0.239
0.203
0.232
0.145
115
USA
Long Beach
0.260
0.211
0.257
0.118
0.260
0.204
0.255
0.108
0.260
0.212
0.256
0.117
116
USA
Los Angeles
0.772
0.548
0.726
0.214
0.822
0.544
0.773
0.136
0.772
0.541
0.725
0.202
117
USA
Miami
0.137
0.098
0.127
0.046
0.137
0.094
0.126
0.034
0.137
0.097
0.129
0.042
118
USA
Mobile
New
Orleans
0.059
0.047
0.058
0.025
0.060
0.047
0.059
0.024
0.059
0.048
0.058
0.025
USA
0.391
0.329
0.366
0.257
0.413
0.318
0.387
0.192
0.391
0.326
0.371
0.255
120
USA
New York
0.207
0.159
0.191
0.108
0.251
0.190
0.231
0.122
0.251
0.194
0.238
0.127
121
USA
0.420
0.335
0.413
0.184
0.420
0.324
0.407
0.171
0.420
0.336
0.411
0.184
102
105
106
107
108
109
110
111
119
USA
Oakland
Port
Everglades
0.209
0.166
0.205
0.086
0.209
0.163
0.206
0.083
0.209
0.167
0.206
0.085
123
USA
Savannah
0.202
0.145
0.191
0.053
0.202
0.141
0.187
0.053
0.202
0.145
0.192
0.054
124
USA
Seattle
0.159
0.122
0.154
0.066
0.159
0.118
0.151
0.060
0.159
0.123
0.153
0.065
125
USA
Tacoma
Puerto
Cabello
0.077
0.058
0.075
0.031
0.077
0.057
0.073
0.027
0.077
0.059
0.074
0.030
0.246
0.199
0.226
0.144
0.395
0.305
0.377
0.162
0.246
0.196
0.231
0.141
122
126
Venezuela
Note
s:
1
37
Table A.3: DEA scores1 under varying assumptions on returns to scale for crude oil terminals

Output: y1=crude oil (DWT)
Inputs: x1=Quay length, x2=Max vessel capacity (dwt), x3=Max draught/depth, x4=Tank storage capacity (m3), x5=Discharge rate (t/h) , x6=Pipeline/loading arm capacity (mm)

Stand
ard
DEA
DMU
7
8
3
2
16
5
1
4
6
13
14
10
28
9
33
11
15
20
27
22
24
25
19
23
Bootstrapped DEA
Stand
ard
DEA
Bootstrapped DEA
Port
Score
Score
2
Upper
value
Lower
value
Score
Score
Upper
value
Lower
value
Score
Score
Upper
value
USA
United
Arab
Emirates
Galveston
1.000
0.730
0.906
0.358
1.000
0.640
0.967
0.099
1.000
0.736
0.907
Jebel Dhanna
Terminal
0.964
0.722
0.902
0.371
1.000
0.758
0.951
0.454
0.964
0.720
0.903
Iran
Saudi
Arabia
Kharg Is.
Juaymah
Terminal
0.963
0.718
0.929
0.311
0.968
0.798
0.949
0.518
0.963
0.718
0.926
1.000
0.691
0.894
0.212
1.000
0.755
0.949
0.403
1.000
0.690
0.874
Qatar
Ras Laffan
Mina al
Ahmadi
0.907
0.687
0.835
0.410
1.000
0.634
0.965
0.000
0.907
0.686
0.833
0.807
0.616
0.781
0.270
0.817
0.685
0.800
0.431
0.807
0.613
0.778
Fujairah
Anch.
1.000
0.599
0.897
0.124
1.000
0.655
0.955
0.103
1.000
0.602
0.890
Ras Tanura
1.000
0.587
0.839
0.066
1.000
0.644
0.964
0.001
1.000
0.583
0.881
Rotterdam
0.748
0.542
0.706
0.200
0.782
0.628
0.752
0.373
0.748
0.540
0.702
Russia
Primorsk
0.664
0.496
0.631
0.239
1.000
0.704
0.968
0.234
0.664
0.500
0.629
USA
South
Korea
Long Beach
0.605
0.480
0.577
0.270
0.605
0.467
0.594
0.225
0.605
0.484
0.580
Yosu
0.602
0.462
0.555
0.293
0.800
0.652
0.771
0.472
0.602
0.464
0.549
USA
0.580
0.452
0.548
0.280
0.711
0.504
0.675
0.104
0.580
0.451
0.545
India
Corpus Christi
Jamnagar
Terminal
0.583
0.436
0.564
0.158
0.590
0.490
0.577
0.294
0.583
0.437
0.565
China
Hong Kong
0.551
0.428
0.519
0.265
1.000
0.638
0.974
0.001
0.551
0.428
0.518
China
Ningbo
0.517
0.404
0.481
0.265
0.646
0.536
0.624
0.417
0.517
0.405
0.479
China
Qingdao
0.628
0.401
0.576
0.078
1.000
0.638
0.960
0.000
0.628
0.401
0.576
Malaysia
South
Korea
Pasir Gudang
0.427
0.338
0.398
0.208
0.429
0.312
0.415
0.117
0.427
0.340
0.400
Daesan
Novorossiysk/
Commercial
Sea Port
0.467
0.328
0.431
0.153
1.000
0.638
0.964
0.023
0.467
0.330
0.435
0.395
0.324
0.382
0.230
0.469
0.403
0.455
0.327
0.395
0.325
0.379
China
South
Korea
Zhoushan
0.390
0.322
0.376
0.224
0.511
0.445
0.496
0.355
0.390
0.323
0.374
Ulsan
0.432
0.319
0.416
0.117
0.437
0.357
0.424
0.216
0.432
0.319
0.415
France
Fos
0.371
0.292
0.352
0.158
0.371
0.276
0.364
0.126
0.371
0.295
0.357
USA
Texas City
0.368
0.290
0.357
0.157
1.000
0.768
0.945
0.469
0.368
0.292
0.357
Bahamas
Freeport
0.365
0.277
0.350
0.138
0.550
0.409
0.542
0.139
0.365
0.279
0.351
Mexico
Cayo Arcas
0.365
0.267
0.345
0.109
0.395
0.316
0.382
0.196
0.365
0.266
0.343
Italy
Trieste
0.278
0.219
0.252
0.151
0.438
0.364
0.423
0.284
0.278
0.220
0.258
Kuwait
United
Arab
Emirates
Saudi
Arabia
Netherlan
ds
Russia
12
Stand
ard
DEA
Bootstrapped DEA

(NIRS)
Country
17
18
38
Lowe
r
value
0.35
6
0.36
8
0.32
4
0.22
6
0.39
6
0.27
0
0.12
5
0.07
0
0.20
8
0.23
1
0.29
0
0.29
3
0.27
2
0.17
0
0.26
1
0.26
9
0.07
8
0.20
2
0.14
5
0.23
1
0.22
9
0.12
4
0.16
9
0.15
3
0.13
2
0.11
2
0.15
4
31
36
32
43
Spain
South
Africa
Algeciras
0.278
0.213
0.270
0.112
0.419
0.320
0.414
0.110
0.278
0.215
0.269
Durban
0.224
0.177
0.218
0.097
0.698
0.547
0.664
0.346
0.224
0.179
0.220
India
Paradip
Jawaharlal
Nehru
New
Mangalore
0.217
0.174
0.203
0.118
0.353
0.295
0.338
0.221
0.217
0.175
0.204
0.226
0.157
0.216
0.062
1.000
0.630
0.959
0.027
0.226
0.157
0.218
0.211
0.153
0.201
0.064
0.798
0.605
0.781
0.280
0.211
0.154
0.201
Taiwan
Saudi
Arabia
Kaohsiung
0.192
0.151
0.183
0.094
0.209
0.178
0.202
0.135
0.192
0.152
0.183
Yanbu
0.209
0.149
0.197
0.050
0.212
0.167
0.205
0.096
0.209
0.149
0.193
Iran
Bandar Abbas
0.182
0.141
0.176
0.073
0.650
0.516
0.630
0.322
0.182
0.143
0.175
France
Le Havre
0.182
0.133
0.169
0.061
0.184
0.146
0.175
0.093
0.182
0.132
0.165
Japan
Kawasaki
0.166
0.132
0.158
0.085
0.183
0.155
0.177
0.117
0.166
0.132
0.156
Spain
Las Palmas
0.172
0.128
0.163
0.062
0.894
0.727
0.869
0.449
0.172
0.129
0.164
Taiwan
Mai-Liao
0.157
0.126
0.150
0.086
0.196
0.168
0.190
0.134
0.157
0.127
0.147
UK
Milford Haven
0.153
0.126
0.147
0.084
0.192
0.168
0.186
0.132
0.153
0.126
0.147
Japan
Mizushima
0.150
0.119
0.137
0.084
0.230
0.196
0.223
0.155
0.150
0.120
0.137
Australia
South
Korea
Brisbane
0.148
0.111
0.140
0.060
0.453
0.351
0.434
0.195
0.148
0.112
0.139
Incheon
0.128
0.102
0.123
0.061
0.186
0.151
0.182
0.082
0.128
0.102
0.123
Sweden
Gothenburg
Visakhapatna
m
0.129
0.100
0.122
0.056
0.138
0.111
0.134
0.070
0.129
0.101
0.124
India
0.131
0.095
0.124
0.050
0.233
0.184
0.225
0.124
0.131
0.095
0.119
Spain
Bilbao
0.104
0.085
0.102
0.052
0.158
0.129
0.156
0.067
0.104
0.086
0.103
Pakistan
Karachi
0.099
0.076
0.094
0.044
0.221
0.170
0.213
0.095
0.099
0.076
0.094
China
Xingang
0.086
0.071
0.083
0.048
0.096
0.084
0.094
0.068
0.086
0.071
0.083
Japan
Yokohama
0.084
0.070
0.081
0.048
0.106
0.093
0.103
0.074
0.084
0.070
0.081
Italy
Genoa
0.095
0.069
0.091
0.023
0.095
0.077
0.093
0.044
0.095
0.069
0.091
India
Chennai
0.052
0.043
0.050
0.029
0.083
0.068
0.080
0.051
0.052
0.043
0.050
Japan
Nagoya
0.044
0.037
0.044
0.026
0.053
0.046
0.051
0.037
0.044
0.038
0.043
USA
Honolulu
0.047
0.035
0.045
0.017
0.243
0.201
0.239
0.123
0.047
0.035
0.046
Romania
Constantza
0.040
0.032
0.039
0.018
0.060
0.047
0.059
0.020
0.040
0.032
0.039
USA
Savannah
0.038
0.026
0.037
0.009
1.000
0.730
0.947
0.353
0.038
0.026
0.037
Italy
South
Korea
Taranto
0.030
0.022
0.028
0.013
0.037
0.030
0.035
0.021
0.030
0.022
0.028
Gwangyang
0.025
0.020
0.024
0.013
0.029
0.024
0.027
0.019
0.025
0.020
0.024
Thailand
Bangkok
0.024
0.016
0.023
0.006
1.000
0.636
0.947
0.000
0.024
0.016
0.023
Japan
0.025
0.016
0.024
0.004
0.025
0.018
0.025
0.004
0.025
0.016
0.024
USA
Chiba
Port
Everglades
0.021
0.014
0.020
0.005
0.442
0.334
0.427
0.184
0.021
0.014
0.020
Indonesia
Jakarta
0.020
0.014
0.019
0.006
1.000
0.632
0.951
0.001
0.020
0.014
0.019
Kenya
Mombasa
0.017
0.012
0.015
0.007
0.023
0.016
0.023
0.001
0.017
0.012
0.015
UK
0.015
0.011
0.014
0.005
1.000
0.642
0.965
0.001
0.015
0.011
0.014
Russia
London
St.
Petersburg
0.012
0.008
0.011
0.004
0.100
0.080
0.097
0.053
0.012
0.008
0.011
Peru
Callao
0.010
0.007
0.009
0.003
0.049
0.041
0.048
0.025
0.010
0.007
0.009
Oman
Salalah
0.008
0.006
0.008
0.003
1.000
0.638
0.953
0.001
0.008
0.006
0.008
Turkey
Ambarli
0.004
0.003
0.004
0.001
0.010
0.007
0.009
0.003
0.004
0.003
0.004
Japan
Kokura
0.001
0.000
0.001
0.000
0.001
0.001
0.001
0.001
0.001
0.000
0.001
Japan
Kobe
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
Malta
Marsaxlokk
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
India
37
India
29
21
39
26
30
41
35
34
38
44
47
50
46
49
51
45
48
40
53
52
55
57
58
54
56
60
42
59
63
61
66
62
64
65
67
68
70
69
Notes
:
1
39
0.10
7
0.10
0
0.12
1
0.05
8
0.06
5
0.09
5
0.05
2
0.07
0
0.06
0
0.08
6
0.05
8
0.08
7
0.08
7
0.08
6
0.06
0
0.06
2
0.05
5
0.05
1
0.05
3
0.04
4
0.04
8
0.04
9
0.02
4
0.02
9
0.02
6
0.01
6
0.01
9
0.00
9
0.01
3
0.01
3
0.00
6
0.00
4
0.00
5
0.00
5
0.00
6
0.00
5
0.00
4
0.00
3
0.00
3
0.00
2
0.00
0
0.00
0
0.00
0
Table A.4: DEA scores1 under varying assumptions on returns to scale for iron-ore terminals

Output: y1=iron ore (DWT)
Inputs: x1=Quay length, x2=Maximum depth (m), x3=Loading/unloading capacity (t/h), x4=Storage capacity (tonnes)

Stand
ard
DEA
DMU
Country
Port

Stand
ard
DEA
Bootstrapped DEA
Bootstrapped DEA

(NIRS)
Stand
ard
DEA
Bootstrapped DEA
Score
Score
2
Upper
value
Lower
value
Score
Score
Upper
value
Lower
value
Score
Score
Upper
value
Lower
value
0.949
0.777
0.937
0.472
1.000
0.828
0.978
0.580
0.949
0.764
0.934
0.460
1.000
0.733
0.946
0.354
1.000
0.783
0.979
0.461
1.000
0.725
0.951
0.337
Brazil
Port
Walcott
Ponta da
Madeira
Brazil
Sepetiba
0.873
0.718
0.861
0.459
0.943
0.812
0.929
0.592
0.873
0.709
0.861
0.435
Australia
Gladstone
0.767
0.642
0.760
0.417
1.000
0.738
0.983
0.182
0.767
0.643
0.759
0.415
Brazil
Tubarao
0.679
0.513
0.662
0.257
0.684
0.563
0.677
0.333
0.684
0.510
0.668
0.237
12
China
South
Africa
Shanghai
Saldanha
Bay
0.600
0.506
0.594
0.334
1.000
0.732
0.979
0.128
0.600
0.506
0.594
0.330
0.587
0.476
0.577
0.289
1.000
0.733
0.979
0.128
0.587
0.475
0.578
0.290
Australia
0.506
0.401
0.496
0.211
0.592
0.509
0.582
0.360
0.506
0.398
0.494
0.206
0.349
0.294
0.346
0.194
1.000
0.735
0.979
0.129
0.349
0.294
0.346
0.189
Canada
Netherland
s
Dampier
Visakhapat
nam
Seven
Islands
Velsen/Ijmu
iden
0.283
0.240
0.281
0.160
1.000
0.736
0.981
0.129
0.283
0.240
0.281
0.158
0.257
0.205
0.252
0.124
0.272
0.230
0.266
0.159
0.257
0.204
0.251
0.122
10
Germany
Hamburg
0.248
0.198
0.243
0.112
1.000
0.739
0.979
0.130
0.248
0.198
0.242
0.117
15
China
Guangzhou
0.195
0.163
0.193
0.106
1.000
0.733
0.980
0.128
0.195
0.163
0.193
0.105
Japan
Kashima
Port
Hedland
0.180
0.139
0.176
0.071
0.250
0.209
0.246
0.146
0.180
0.138
0.176
0.072
0.103
0.080
0.100
0.041
0.128
0.110
0.126
0.078
0.103
0.079
0.100
0.041
Australia
6
4
13
India
14
8
11
Australia
Notes:
40
Table A.5: DEA scores1 under varying assumptions on returns to scale for coal bulk terminals

Output: y1=coal (DWT)
Inputs: x1=Quay length, x2=Storage capacity (tonnes), x3=Loading/unloading (total capacity per h)

Stand
ard
DEA
DMU
Bootstrapped DEA

Stand
ard
DEA
Bootstrapped DEA

(NIRS)
Stand
ard
DEA
Bootstrapped DEA
Country
Port
Score
Score
2
Upper
value
Lower
value
Score
Score
Upper
value
Lower
value
Score
Score
Upper
value
Lower
value
Australia
Netherland
s
Hay Point
Velsen/Ijm
uiden
Banjarmar
sin
1.000
0.755
0.925
0.426
1.000
0.718
0.928
0.392
1.000
0.724
0.890
0.393
1.000
0.676
0.888
0.248
1.000
0.667
0.893
0.175
1.000
0.678
0.896
0.259
0.969
0.670
0.853
0.265
0.969
0.644
0.874
0.167
0.969
0.670
0.867
0.276
0.948
0.669
0.887
0.269
1.000
0.593
0.900
0.046
0.948
0.676
0.894
0.272
China
Shanghai
Qinhuang
dao
1.000
0.664
0.903
0.241
1.000
0.643
0.875
0.229
1.000
0.642
0.887
0.228
Australia
Newcastle
1.000
0.660
0.895
0.259
1.000
0.646
0.905
0.240
1.000
0.643
0.895
0.242
China
Rizhao
Guangzho
u
1.000
0.660
0.859
0.208
1.000
0.662
0.901
0.208
1.000
0.667
0.873
0.218
1.000
0.657
0.855
0.234
1.000
0.595
0.912
0.045
1.000
0.655
0.890
0.236
1.000
0.628
0.866
0.174
1.000
0.594
0.908
0.046
1.000
0.628
0.887
0.173
3
9
11
10
2
12
7
Indonesia
China
China
Colombia
Gladstone
Puerto
Bolivar
0.917
0.598
0.842
0.173
1.000
0.594
0.894
0.045
0.917
0.603
0.864
0.168
16
Japan
Oita
0.541
0.382
0.489
0.154
0.541
0.352
0.504
0.070
0.541
0.379
0.497
0.161
Brazil
Sepetiba
0.477
0.363
0.447
0.213
0.676
0.520
0.644
0.310
0.676
0.516
0.639
0.304
23
Italy
South
Africa
Taranto
Richards
Bay
0.419
0.317
0.395
0.160
1.000
0.595
0.891
0.046
0.419
0.309
0.388
0.159
0.446
0.291
0.399
0.095
0.576
0.389
0.526
0.150
0.576
0.386
0.525
0.151
India
Paradip
0.362
0.265
0.328
0.145
0.646
0.507
0.620
0.303
0.646
0.503
0.616
0.297
15
Canada
Vancouver
0.280
0.216
0.259
0.129
0.281
0.206
0.259
0.116
0.280
0.207
0.259
0.119
27
Egypt
Alexandria
0.256
0.186
0.240
0.082
1.000
0.597
0.892
0.046
0.256
0.182
0.234
0.084
19
Japan
0.244
0.172
0.229
0.064
0.294
0.197
0.265
0.060
0.244
0.175
0.231
0.068
0.192
0.139
0.178
0.064
0.205
0.145
0.191
0.067
0.205
0.145
0.187
0.067
0.183
0.130
0.165
0.060
0.204
0.144
0.186
0.069
0.204
0.143
0.180
0.067
0.191
0.129
0.177
0.044
0.205
0.137
0.183
0.044
0.191
0.131
0.179
0.047
0.132
0.093
0.120
0.044
0.239
0.182
0.226
0.105
0.239
0.180
0.225
0.102
13
Australia
Indonesia
Nagoya
Port
Kembla
Abbot
Point
Amsterda
m
Tanjung
Bara
Balikpapa
n
0.130
0.092
0.119
0.038
0.146
0.100
0.130
0.044
0.146
0.103
0.130
0.047
22
USA
Mobile
0.091
0.069
0.085
0.039
0.099
0.073
0.092
0.041
0.099
0.073
0.091
0.040
21
USA
Baltimore
Long
Beach
0.091
0.066
0.086
0.028
0.109
0.079
0.101
0.039
0.109
0.078
0.100
0.036
0.088
0.061
0.083
0.023
0.091
0.060
0.082
0.017
0.088
0.062
0.083
0.024
Tarragona
Hampton
Roads/Nor
folk
0.082
0.055
0.076
0.018
0.082
0.056
0.078
0.018
0.082
0.056
0.077
0.019
0.068
0.050
0.064
0.024
0.079
0.057
0.071
0.029
0.079
0.057
0.071
0.028
0.059
0.043
0.053
0.022
0.077
0.049
0.071
0.006
0.059
0.043
0.053
0.022
Poland
La Spezia
Swinoujsci
e
0.057
0.042
0.052
0.023
0.060
0.043
0.054
0.023
0.060
0.044
0.054
0.024
Australia
Brisbane
0.049
0.037
0.044
0.021
0.049
0.035
0.044
0.019
0.049
0.035
0.044
0.019
18
17
20
14
25
29
26
Australia
Australia
Netherland
s
Indonesia
USA
Spain
24
USA
33
30
31
Italy
41
Canada
Prince
Rupert
0.046
0.033
0.043
0.014
0.058
0.042
0.054
0.020
0.058
0.041
0.054
0.019
32
Poland
Gdansk
0.031
0.023
0.028
0.014
0.043
0.033
0.040
0.020
0.043
0.033
0.040
0.020
34
Estonia
Tallinn
0.012
0.008
0.011
0.003
1.000
0.596
0.907
0.046
0.012
0.009
0.011
0.004
28
Notes:
42
Table A.6: DEA scores1 under varying assumptions on returns to scale for grain bulk terminals

Output: y1= grains (DWT)
Inputs: x1=Quay length, x2=Storage capacity (tonnes), x3=Loading/unloading (total capacity per h)

Standa
rd DEA
DMU

Stand
ard
DEA
Bootstrapped DEA
Stand
ard
DEA
Bootstrapped DEA
Uppe
r
value
Lowe
r
value
Score
Score
Uppe
r
value
Lowe
r
value
Score
Score
Uppe
r
value
Lowe
r
value
Country
Port
Score
Japan
Chiba
San LorenzSan Martin
Bahia
Blanca
1.000
0.583
0.781
0.079
1.000
0.592
0.805
0.077
1.000
0.608
0.831
0.109
1.000
0.551
0.689
0.056
1.000
0.539
0.682
0.010
1.000
0.556
0.762
0.053
1.000
0.567
0.693
0.072
1.000
0.540
0.706
0.010
1.000
0.584
0.760
0.094
USA
South
Korea
Duluth
1.000
0.569
0.676
0.078
1.000
0.539
0.701
0.010
1.000
0.579
0.771
0.085
Busan
0.960
0.555
0.732
0.087
1.000
0.539
0.671
0.010
0.960
0.566
0.817
0.097
Argentina
Rosario
0.185
0.124
0.154
0.044
0.455
0.339
0.434
0.146
0.455
0.351
0.437
0.176
Brazil
Rio Grande
0.383
0.249
0.310
0.079
0.463
0.292
0.378
0.090
0.463
0.304
0.398
0.115
Brazil
South
Korea
Tubarao
1.000
0.608
0.771
0.134
1.000
0.541
0.730
0.010
1.000
0.628
0.831
0.171
Incheon
0.271
0.166
0.227
0.037
0.338
0.204
0.269
0.044
0.338
0.213
0.286
0.060
10
Japan
Mizushima
0.213
0.139
0.166
0.045
0.327
0.220
0.268
0.084
0.327
0.228
0.281
0.101
11
Taiwan
Taichung
0.209
0.137
0.168
0.046
0.321
0.217
0.268
0.085
0.321
0.225
0.283
0.102
12
China
Shanghai
0.243
0.150
0.177
0.039
0.343
0.215
0.269
0.065
0.343
0.224
0.289
0.084
13
Canada
Vancouver
0.649
0.417
0.510
0.127
1.000
0.541
0.667
0.010
0.649
0.414
0.501
0.142
14
Japan
Kashima
0.275
0.166
0.213
0.030
0.275
0.166
0.210
0.026
0.275
0.174
0.227
0.041
15
USA
Houston
0.169
0.109
0.139
0.033
0.187
0.115
0.145
0.031
0.187
0.120
0.156
0.041
16
USA
Mobile
0.402
0.247
0.307
0.061
1.000
0.538
0.673
0.010
0.402
0.254
0.328
0.072
17
Canada
Montreal
0.063
0.042
0.053
0.015
0.146
0.108
0.139
0.046
0.146
0.112
0.140
0.055
18
Canada
Thunder Bay
0.555
0.348
0.503
0.079
1.000
0.540
0.702
0.010
0.555
0.353
0.503
0.089
19
Brazil
Santos
0.083
0.052
0.067
0.014
0.131
0.084
0.104
0.027
0.131
0.087
0.113
0.033
20
Kuwait
Netherlan
d
Kuwait
0.099
0.064
0.083
0.019
0.125
0.080
0.106
0.024
0.125
0.083
0.112
0.031
Rotterdam
0.119
0.072
0.101
0.010
0.119
0.071
0.099
0.009
0.119
0.074
0.104
0.012
22
France
Rouen
0.409
0.258
0.372
0.059
0.736
0.416
0.605
0.017
0.409
0.263
0.377
0.067
23
Australia
0.050
0.030
0.039
0.007
0.101
0.064
0.084
0.020
0.101
0.067
0.087
0.025
1
2
3
4
5
21
Argentina
Argentina
Score
Bootstrapped DEA

(NIRS)
Canada
Geraldton
Baie
Comeau
0.035
0.023
0.029
0.007
0.073
0.050
0.065
0.021
0.073
0.052
0.067
0.024
25
USA
Beaumont
0.042
0.025
0.033
0.005
0.073
0.045
0.058
0.011
0.073
0.047
0.063
0.015
26
Belgium
Ghent
0.035
0.023
0.028
0.007
0.055
0.038
0.048
0.014
0.055
0.039
0.049
0.017
27
Israel
Haifa
0.020
0.013
0.017
0.003
0.047
0.033
0.044
0.013
0.047
0.034
0.045
0.015
28
Australia
Port Kembla
0.030
0.018
0.023
0.004
0.043
0.026
0.034
0.006
0.043
0.027
0.037
0.007
29
Australia
Freemantle
0.061
0.037
0.047
0.008
0.061
0.036
0.048
0.005
0.061
0.039
0.050
0.010
30
China
Tianjin
0.033
0.019
0.026
0.003
0.033
0.019
0.026
0.002
0.033
0.020
0.028
0.004
31
Australia
Newcastle
Lake
Charles
0.020
0.013
0.017
0.005
0.028
0.019
0.024
0.007
0.028
0.019
0.026
0.008
0.036
0.023
0.033
0.005
0.038
0.023
0.030
0.004
0.036
0.023
0.031
0.006
24
32
USA
43
33
Japan
South
Africa
Hachinohe
0.053
0.032
0.042
0.007
0.053
0.030
0.043
0.003
0.053
0.033
0.046
0.009
East London
0.011
0.007
0.009
0.001
0.025
0.016
0.021
0.005
0.025
0.017
0.023
0.006
35
Egypt
Port Said
0.046
0.029
0.042
0.007
0.054
0.032
0.043
0.006
0.046
0.029
0.042
0.007
36
Australia
Brisbane
0.009
0.006
0.008
0.002
0.020
0.014
0.018
0.006
0.020
0.014
0.019
0.007
37
Spain
Valencia
Buenos
Aires
0.009
0.005
0.006
0.001
0.017
0.010
0.014
0.002
0.017
0.011
0.015
0.003
0.014
0.008
0.012
0.001
0.014
0.008
0.011
0.001
0.014
0.009
0.012
0.001
0.011
0.007
0.008
0.002
0.013
0.009
0.011
0.003
0.013
0.009
0.011
0.003
UK
Portland
Southampto
n
0.005
0.003
0.004
0.001
0.011
0.008
0.010
0.003
0.011
0.008
0.010
0.004
Italy
Trieste
0.001
0.000
0.001
0.000
0.001
0.001
0.001
0.000
0.001
0.001
0.001
0.001
34
38
Argentina
39
Australia
40
41
Notes
:
1
2
44
NOTES
However, according to the review by Trujillo and Gonzales (2008) there are about an equal number of
studies exploring efficiency via estimating a stochastic frontier production with a predefined functional
form, suggesting the absence of consensus vis--vis the best approach to be used.
Cheon, et al., 2010; Wu and Goh, 2010; Martinez-Budria, et al., 2009; Wang and Cullinane, 2006; AlEraqui, et al., 2007; Tongzon, 2001
This mainly legitimates stochastic frontiers and econometrics approaches though they impose a functional
form to the production.
Bootstrapping is a re-sampling method consists in constructing a number of resamples of the observed

dataset, and of equal size, where each of these is obtained by random sampling with replacement from the
original dataset.
45

Efficiency of Container and Bulk Cargo Ports

Uploaded by

Copyright:

Available Formats

Efficiency of Container and Bulk Cargo Ports

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Efficiency of Container and Bulk Cargo Ports

Uploaded by

Copyright:

Available Formats

she'd be free for lunch from 12:45pm-2:30pm or anytime between 4pm-6pm.

Efficiency of world ports in container and

Please cite this paper as:

Merk, O., Dang, T. (2012) Efficiency of world ports in

OECD Regional Development Working

OECD REGIONAL DEVELOPMENT WORKING PAPERS

--------------------------------------------------------------------------OECD Regional Development Working Papers are published on

2. THE METHODOLOGICAL APPROACH TO MEASURE PORT EFFICIENCY

3.1. Container terminals

TEUs 2009 Output May

Source: OECD database

Overall efficiency scores

Source: Authors own calculations.

Jawaharlal Nehru (Ind)

Source: Authors own calculations.

Source: Authors own calculations.

3.2. Oil terminals

Oil terminal sample

Output May Quay length Max vessel

Source: OECD database.

Overall efficiency scores

Source: Authors own calculations.

Source: Authors own calculations.

Source: Authors own calculations.

3.3. Coal bulk terminals

Output May 2011

Source: OECD database.

Quay length Storage capacity Loading/unloa

Overall efficiency scores

Hay Point (Aus)

Source: Authors own calculations.

Efficiency potential gains

Prince Rupert (Can)

Hampton Roads/Norfolk (USA)

Long Beach (USA)

Tanjung Bara (Ind)

Abbot Point (Aus)

Port Kembla (Aus)

Richards Bay (Sou)

Puerto Bolivar (Col)

Hay Point (Aus)

Source: Authors own calculations.

3.4. Iron bulk terminals

Source: OECD database.

Quay Maximum Loading/u

Overall efficiency scores

Port Hedland (Aus)

Seven Islands (Can)

Saldanha Bay (Sou)

Ponta da Madeira (Bra)

Port Walcott (Aus)

Source: Authors own calculations.

Efficiency potential gains

Port Hedland (Aus)

Seven Islands (Can)

Saldanha Bay (Sou)

Ponta da Madeira (Bra)

Port Walcott (Aus)

Source: Authors own calculations.

Output ranking Port names