The measurement of port efficiency using data envelopment analysis

Full text

1
Proceedings of the 9th World Conference on Transport Research, 2001
Topic Area:
A2 Maritime Transport and Ports
Paper Number:
1203
Authors:
V.F.VALENTINE and R.GRAY
Title: The measurement of port efficiency using data envelopment analysis
V.F.Valentine and R.Gray
Institute of Marine Studies, University of Plymouth, Drake Circus, Plymouth, PL4 8AA
United Kingdom
Email: vvalentine@plymouth.ac.uk and rgray@plymouth.ac.uk
Telephone: +44 (0) 1752 232465
Fax: +44 (0) 1752 232406
ABSTRACT
Understanding performance is a concept fundamental to any business, whether it is the
measuring of achievements against set goals and objectives or, against the competition.
Ports are no exception and it is only by comparison that performance can be evaluated.
Ports are, however, a complex business with many different sources of inputs and
outputs which makes direct comparison among apparently homogeneous ports seem
difficult. The subject is further complicated by the various types of port ownership and
organisational structures that exist throughout the world.
This paper investigates the efficiency of differently owned container ports, comparing
privately owned ports against those remaining in the public sector, and those that have
elements of both public and private ownership patterns. In addition, the organisational
structure will be analysed and classified. The paper seeks to determine whether there is
a particular type of ownership and organisational structure that leads to a more efficient
port. The results of this paper should assist governments, port administrators and port
owners in determining the different ways they can structure their ports.
Key words: ‘Port’, ‘Container’, ‘Privatisation’ and ‘Data Envelopment Analysis’
Method of Presentation:
(1) OHP ( X )
(2) Slide Projector ( )
(3) LCD Projector ( )

2
1. INTRODUCTION
Understanding performance is a concept fundamental to any business, whether it is the
measuring of achievements against set goals and objectives or, against the competition.
Ports are no exception and it is only by comparison that performance can be evaluated.
Ports are, however, a complex business with many different sources of inputs and
outputs which makes direct comparison among apparently homogeneous ports seem
difficult. The subject is further complicated by the various types of port ownership and
organisational structures that exist throughout the world. During the last two decades
the ownership of one of the most important trade entry points into any country, the
seaport, has changed from being solely in the hands of national or local governments
into, either wholly or partially, private hands. It is this change, which is called
privatisation, that has attracted much interest from both academics and those working
within the industry. This paper will look at how seaports are owned and how their
structure is organised to determine whether these factors have any relation to its
performance. This paper will use a technique called Data Envelopment Analysis (DEA)
to assess the relative efficiency of a sample of ports and cluster analysis to examine
their organisational structures. The results of this paper will help serve as a guide for
governments, port administrators and port owners on the different ways in which they
can structure their ports to lead to greater efficiency.
2. BACKGROUND
Privatisation is a concept rather than an actual definable process. The word came into
being during the late 1960s and was later attributed to the UK government's reforms to
ownership and operation of numerous companies managed by the state. Chapman
(1990) has accredited Drucker (1969) as the author of the word 'privatization' in its
American spelling. The actual process of implementing privatisation is not however a
new concept. Neither can it be said to have originated in the UK. It was rather a
christening of an established process, a renaissance of an earlier idea on the ownership
and management of a company. What can be said is that the extent to which the UK
government pursued this course of action certainly attracted attention from other
countries which no doubt contributed to the sudden global desire to privatise during the
1980s. A comprehensive review of privatisation methods is given in Abdel-Fattah et al
(1999).
Privatisation in developing countries is often the first phase in a process of industrial
liberalisation and a move towards industrial progression. Viewed as this first step
towards creating free trade it has therefore not surprisingly been a high priority for
developing countries. It begins with the transfer of absolute control of industry away
from the government to private partners with particular expertise. The reasons for this
change are numerous but can be summarised as follows: improvements in efficiency
through private sector management skills; enhancement of service quality through
improved commercial responsiveness; reduction in the fiscal burden of loss-making
state enterprises or the need for the future subsidy; a reduction in the fiscal demands on
central and local government through access to private sector capital; and additional
revenue streams (Port Development International, March 1999).
The value of world-wide privatisations in 1999 grew by 10% over the preceding year
providing governments with US$145 billion (Washington Times 2000). Some countries
have rapidly progressed towards the goal of privatisation whilst others have been

3
hindered by political, fiscal or labour influences, or a general unwillingness to accept
change.
One of the ‘problems’ with privatisation is the perception that it reduces jobs. Indeed,
this is often the immediate reality in many industries that are privatised, as companies
no longer have to accept restrictive employment practices. The longer term view that it
creates efficiency, profitability and growth is not an issue considered by those faced
with the prospect of redundancy.
3. OWNERSHIP STRUCTURE
Cass (1996) in his study of world port privatisation concluded that there were only
really three types of port ownership: public, private or joint public/private. He points
out that the most common type of port privatisation are (1) the sale of operating
concessions, (2) joint public/private ventures, (3) private orientated but port authority
controlled operating subsidiaries, (4) the ‘corporatisation’ of government port agencies
or (5) the dissolution of government owned cargo handling monopolies. The ‘lock,
stock and barrel’ approach of Great Britain and New Zealand are the exceptions. The
degree of public involvement is naturally dependent upon national ideology. Cass
(1996) and Heikkila (1990) both state the examples of the United States where the
municipal authority plays a major part in the operation of the port. There ports compete
against other ports along the coast for business. However, at the other end of the scale
is Taiwan where the administration of the ports is centralised.
Boardman and Vinning (1989) found that certain types of ownership structure the state
owned enterprises and mixed ownership entities performed substantially worse than
similar private companies. They concluded that there were performance differences
between public and private companies in competitive environments and, that where
there was a partial privatisation, the performance was sometimes worse. They claimed
that conflicting ideologies between the two different owners cause what they term
‘cognitive dissonance’. However, Bos (1991) looked at what Tandon (1997) called “the
survey of all the surveys” on the efficiency of public and private firms and came to the
opinion that Boardman and Vinning (1989) had directly opposing views from a
previous study by Borcherding et al (1982). Tandon’s (1997) explanation of these
apparent conflicting views relies not upon the ownership structure but upon the market
conditions in which they operate. Private firms are likely to be in a more competitive
environment and thus more in tune with the need to be efficient than public enterprises
that perhaps operate in a restrictive environment. He argues that in studies involving
public and private firms in the same business, such as airlines, some private airlines are
more profitable but on balance it is approximately equal. This research aims to see
whether this is the case for ports.
Caves et al (1982) in looking at United States private railways and Canadian public
railways concluded that the Canadian public firm was more efficient. Tandon (1997)
states that the process of identifying which approach is more efficient depends upon
disentangling ownership from the effects of deregulation and competition. De Alessi
(1980) states that not only are government firms less efficient but are also less
successful in satisfying the consumer’s needs. Everett and Robinson (1998) in their
research into Australian port reform suggest that the “corporatization” of some ports
has not resulted in the liberalisation and the near private performance that was

4
anticipated. Frech (1980) in looking at the role of property rights within the firm
suggests that if the ownership structure is attenuated this leads to lower firm wealth and
more non-pecuniary benefits. Thus, privatisation, by shortening the ownership structure
should have an opposing effect. Likewise the organisational structure should also play a
significant role by suggesting that simple structures be inherently more efficient than
the more complex machine bureaucracy and divisional structures.
4. ORGANISATIONAL STRUCTURE
Roe (1999) in looking at the newly privatised subsidiaries of the state owned Polish
Ocean Lines observed that there was a desire to avoid control from the parent company
and to change the organisational structure soon after privatisation. Mintzberg (1979)
looked at organisational structures and reached the conclusion that there are essentially
five different types of organisational structure. They are called simple structure,
machine bureaucracy, professional bureaucracy, divisional structure and adhocracy. As
far as ports are concerned only three of these seem to fit into the modern day port
structure. First let us consider the options that do not fit. The adhocracy does not fit into
the structure of any port because of its lack of rigidity. Suitable for software companies
and film producers, its role within a port would likely lead to chaos. Ports require
careful planning and development based upon what may be needed 10 or 20 years into
the future. Without the rigidity of a formal structure each element in the chain would
not know the whole picture, only the person at the top may see everything. Likewise
the professional bureaucracy is not suitable in a port because of the routine and
repetitive tasks that are commonplace within a port’s day to day service. The
professional bureaucracy is typical of industries that require highly professional people
to perform routine tasks in an unsupervised manner such as solicitors and accountants.
Whilst professional people are required in certain areas of port activity, and qualified
personnel are needed to operate expensive and dangerous machinery, a professional
bureaucracy would not be appropriate. This leaves us with the three remaining
structures that are prevalent in the port industry, viz. simple structure, machine
bureaucracy and divisional structure.
The simple structure is the most flexible, allowing separate divisions/departments
reporting straight to the top decision-maker. As the name suggests it is usually the first
stage through which a company progresses in its evolution. This structure by its
simplicity is therefore likely to be the most efficient.
The machine bureaucracy is characterised by its many departments reporting up a chain
of command to a line manager before reporting to the top decision-maker. Because the
decision making has to follow a long process before it reaches the top, decisions tend to
be slower. These structures tend to be found in government owned enterprises and
hence the inclusion of port bodies and corporatisation in this category.
The divisional structure occurs when companies operate within large areas. Each
department has to report to a regional office that in turn reports to a select group of
managers before information is passed to the top decision-maker. This structure can be
best seen in the municipal ports of the UK and the port societies of Columbia. These
divisional structures tend to operate where there are joint public/private enterprises or
where conglomerates own the port.

5
5. METHODS OF ANALYSIS
For the purpose of this research data was collected from a sample of thirty one
container ports from the top 100 container ports as published annually by Cargo
Systems journal. The data relates to 1998, this being the latest available at the time this
research was conducted.
5.1. Measuring Efficiency
Table 1 shows the raw data that will be used to measure the relative efficiency of the
sample ports. The method adopted for this analysis is Data Envelopment Analysis
(DEA). DEA is an established statistical technique that measures the relative
efficiencies of units where simple efficiency measures are difficult to obtain (Farrell
1957 and Charnes et al 1978). The main attraction of DEA is that it can deal with
multiple inputs and outputs. The units in any DEA assessment are generally
homogeneous and independent units performing the same function, and it is of most
use where there are a large number of units providing an ‘identical’ service in relative
isolation (Szczepura et al 1992). DEA was first developed as a way of measuring
service units by Charnes et al (1978) and was based upon Farell’s (1957) idea of
linking the estimation of technical efficiency and production frontiers. The model has
since been added to and developed over the years. Between 1978 and 1992 over 400
articles, books and dissertations were published on DEA (Charnes et al 1995).
Warwick Business School has pioneered the research and is regarded as one of the
leading institutions working in this field. DEA has been successfully used to research
airports (Gillen and Lall, 1997 and De La Cruz 1999), local government authorities,
courts, hospitals, general medical practitioners and bank branches to test efficiency
where there are multiple centres of inputs and outputs. Its application to the port
industry would therefore appear to be ideal. There have however only been a few
studies involving seaports using DEA. Martinez-Budria et al (1999) and Tongzon
(2001) are two studies using Spanish and Australian ports, respectively. Roll and
Hayuth (1993) in a hypothetical study state that DEA is a most suitable tool for
measuring port efficiency.
Sachis (1996) looked at the different techniques for measuring productivity and
confirmed DEA’s usefulness. However his research adopted an engineering method to
take account of the technological investments when looking at the efficiency of Israeli
ports. Various other studies have used the assessment of productivity based upon output
per worker (DeMonie 1987) or output per wharf (Frankel 1991), whilst others use
production functions (Kim and Sachish 1986, DeNeufville and Tsunokawa 1981).
Gillen and Lall (1997) looked at airport terminals and chose two outputs, number of
passengers and pounds weight of cargo. They chose six inputs: number of runways,
number of gates, terminal area, number of employees, number of baggage collection
belts and number of public parking places. They conclude that the number of gates has
the greatest overall effect upon efficiency. In terms of ports, gates, which facilitate the
loading of the cargo, could be equated to loading cranes, and runways to berths.
Efficiency can simply be expressed as a ratio of output to input provided that the
product only produces one output. However, as most institutions produce multiple
outputs from multiple inputs each variable must be given a weighting to produce a
more accurate result. Efficiency then begins to resemble the sum of weighted outputs
over the sum of weighted inputs. As the method of weighting can be biased towards

6
any particular outcome, the DEA technique allows for each weighted input/output to be
seen in its most favourable light. A greater number of variables entered into the formula
means less emphasis on any particular piece of data. Therefore Szczepura et al (1992)
argues that the number of variables should be kept to as low as possible. Weighting has
also been omitted within this calculation to reduced the possibility of ambiguous
results.
5.2 Analysing Organisational Charts
Once the organisation charts of the selected ports had been collated it was necessary to
classify the ports into the three organisational types as identified by Mintzberg (1979)
and discussed above. The variety of sizes and styles of organisational chart made the
identification process arduous but not impossible. Mintzberg (1979) identified several
different organisational structures that could be applied to any business. The three
structures identified earlier as being relevant to ports were simple, divisional and
machine bureaucracy structures. The simple structure is simply two layers, the
divisional structure “short and narrow” and the machine bureaucracy “tall and fat”.
When it came to applying these three distinct types of organisation to the actual
organisational of ports several, or indeed most, did not appear to fit. Therefore it was
necessary to convert each chart into a numerical format in order that it could be
analysed on a more objective basis. Because each structure as identified by Mintzberg
(1979) has a characteristic shape to it, it was decided that the most accurate method of
equating the organisation model to a formula was to give numeric values to the number
of units, divisions and layers. The number of units represented each activity within the
organisation. Each activity is the process, responsibility, function or undertaking that is
being generated such as the maintenance or accounting departments. All companies
consist of such functions which combined represent the entire activity of the business.
The divisions within an organisation chart were determined by the number of distinct
vertical linear pathways. Where there was little or no interaction between these distinct
vertical linear pathways it could be evidenced that these were separate divisions that
operated on a vertical reporting structure characteristic of a divisional organisation. In
this type of organisation, because of the vertical reporting structure, it is sometimes
commonplace for departments working in parallel not to be aware of what each other is
doing. In terms of ports, divisions such as technical and administration and planning
could be present within a large port or the divisions may occur where several ports are
grouped together under one ownership. The next characteristic to be analysed
numerically was the linear stratification amplitude or the number of layers apparent
within the organisation structure. Layers traditional start at the top with the board of
directors and down through the chairman, human resources, etc.
5.3 Cluster Analysis
Cluster analysis is one of a group of multivariate techniques. As the name implies
cluster analysis is a statistical technique which sets out to solve problems in data by
grouping together similar individuals or objects into clusters. The members of the
clusters are more alike to each other than members of other clusters. Thus the aim of
cluster analysis is to interpret the variables by placing them into new groups that can be
easily understood (Aaker et al 1995). It is used in a multitude of disciplines such as
psychology, biology, sociology, economics, engineering and business (Hair et al 1995).
It is similar to factor analysis but whilst factor analysis is concerned with grouping

7
variables, cluster analysis groups objects together. It is particularly useful as a data
reduction technique when there is a large amount of data that would prove arduous to
interpret. By grouping the results into clusters or groups it is then easier to understand
what each group represents and thus its place in the overall picture. Whilst it is
commonly used in an exploratory role it can also be used to test hypotheses by
comparing the results with the researcher’s expectations. As with factor analysis,
cluster analysis is not a statistical inference technique where parameters from a sample
are said to be indicative of the entire population. On the contrary, cluster analysis is an
objective methodology for quantifying the structural characteristics of a set of
observations. Hair et al (1995) state that care should be used when relying upon the
results of cluster analysis because of the degree of input and interpretation need by the
researcher. They state that cluster analysis is more of an art than a science but used
cautiously the results can be very informative.
The process begins by selecting the variables to be included into the process. It is then
the researcher’s choice as to how many clusters are included. These two points are
hence fundamental in the final results. The actual variables selected in this research are
four and the required number of clusters was three. The results of the analysis showed
that although three distinct clusters could be shown, two distinct clusters were evident.
5.4 Discriminant analysis
Discriminant analysis is similar to cluster analysis in that it attempts to divide
individuals and not variables into distinct groups. Discriminant analysis differs from
cluster analysis in that the number of characteristics derived from the clusters are not
usually known before the analysis and the group numbers and membership are
predicted. Discriminant analysis creates a regression equation that uses a dependent
variable that is discrete rather than continuous (George and Mallery 1999). It uses pre-
existing data in which group membership is already known, a regression equation can
then be calculated that discriminates between the groups. This pre-existing data comes
from previous studies for which the groups and results are already known.
6. ANALYSING THE RESULTS
Ports are a complex business with many different sources of inputs and outputs. Van
Niekerk (2000) states that, because of the complexity of ports, not even the inputs and
outputs can be easily defined and applied to seemingly homogeneous ports. However
this problem can to some extent be overcome by using multiple inputs and outputs. As
mentioned earlier, there are two methods of analysis in this research, DEA that relates
to port efficiency and cluster analysis to organisational structure.
For the purpose of this research the output variables to measure the efficiency of a port
are container throughput and total throughput (i.e. the quantity of goods transported by
the port from which it generates its main income). Container throughput is used
because it is the primary source of comparison between container ports. The reason
that the total throughput figure is being used is because it gives the exact output of the
port as a single homogeneous figure that represents all of the cargo transported. It is
also a figure used by all ports to measure the level of business transacted. It means that
the number of containers and amount of bulk cargo transhipped are converted to a
universally accepted cargo throughput figure to enable comparisons of size. Marx
(1970) in his treatise on the nature of class and its relationship to the means of

8
production listed three fundamental aspects of modern capitalism, which are land,
labour and capital. These three elements represent the essential components of modern
companies. Thus in order to compare companies it makes sense to analyse these
ubiquitous elements. In terms of ports land can be said to be the area owned by the port,
labour the number of employees and capital the net assets of the port. Thus bearing this
in mind, it was originally thought necessary to include profit and the net assets of the
port into the equation, since these are the end factors which all ports seek. However, a
preliminary study conducted by one of the authors found that these figures were
invariably only available for ports that were publicly owned or publicly listed on a
stock market. Manpower, being second on the list, was also a factor originally thought
worthy of inclusion. As ports become privatised they tend to reduce their number of
employees to more profitable levels. However, with some ports that were owned by
joint public and private companies the information was difficult to ascertain. A port
authority may openly state that it employs, say 100 people, but within the port there
could be several terminals each employing another 100 persons, thus making an unfair
comparison between some ports. For example the port of Melbourne with a throughput
of around 42mt states that it employs 81 personnel, and Mumbai which has a
throughput of 30mt claims to employ over 24,000 personnel. Clearly such a
discrepancy between these two figures reveals that there must be a difference in the
way the figures are computed since they suggest that it takes 300 people in India to do
the same a one person in Australia. Although it could be argued that there may be a
higher use of equipment in the more industrial Australia this factor alone would seem
an inadequate explanation. Next on Marx’s list is land; in the case of ports the total area
of the ports was calculated differently by different ports. Some ports chose to include
the area of water encompassing the port’s boundaries whilst others only measured
actual land. Thus, in order to overcome this lack of uniformity of measurement within
the industry in calculating land and labour, the authors decided to use quay length (both
container and overall quay length). Thus in summary the input of length of container
quay will be compared against the output of container throughput and total quayage
against total throughput.
6.1 Results from the Application of Data Envelopment Analysis
The results of table 2 relate to DEA and show that by applying this analysis three ports
are considered to be 100% efficient. The results show that the two most efficient ports
are also the world ranking number one and number two ports in terms of container
throughput. The inclusion of Santos as holder of joint first place shows that this port for
its size is relatively efficient.
The calculations are capped at 100% since this is the maximum efficiency that can be
reached. If however this cap is removed then the relative efficiency rating for the ports
are Santos, 145.22, Singapore 166.72 and Hong Kong 181.12. This means that Hong
Kong is the most efficient container port examined in this study. The gap between the
so-called top 3 who scored 100% and that of fourth position, Houston, is quite
considerable. Houston has a relative efficiency rating of 69.86, is approximately half
that of Santos. This difference may result from the main priorities of the top three ports
being container traffic, whereas the other ports in the study seem to diversify into other
areas. The valued added through adopting this method of research is that ports are
analysed for their relative efficiency. Therefore with more ports added or subtracted,

9
the relative efficiency will change. Ports in certain regions can also be compared with
those from another, helping to highlight regional differences that may exist.
6.2 Results from the Application of Cluster Analysis
Table 3 shows the numeric values awarded to each port in terms of the above
mentioned three distinct organisational criteria, viz. units, divisions and layers. Thus,
we can see in the second column that if cluster analysis is required to provide three
distinct clusters then all those ports appearing in cluster one are represented in column
two with the number ‘1’. Similarly, those falling into clusters two and three are also
marked with the number ‘2’ or ‘3’. Column three provides the results if only two
clusters are required. It can be seen that the distinction between clusters two and three
is not as strong as between one and two.
The next three columns show how many units, divisions and layers each of the thirty-
one ports have present. Thus, reading from the top of the list we can see that the port of
Antwerp has ten operational units, three divisions and three layers. Compare this with
Auckland, the first port alphabetically in cluster two, which has 30 operational units,
five divisions and nine layers.
If however we look at the characteristics of each cluster we can see that the ports in
cluster one have an overall average number of fourteen units, six divisions and five
layers. In comparison with the other two clusters we can determine that the
characteristics of cluster one are that it has few units, a medium horizontal range of
division and a short vertical range of layers which is reminiscent of the simple
structure. This differs substantially from the “ideal” example given by Mintzberg
(1979) of a simple structure two layers deep and elongated horizontally. It shows that
the structure of a port can contain several layers and divisions but the key element of
this cluster is that it must have very few functional units
In cluster two the average number of units is almost two and a half times that of cluster
one at thirty four units. The average number of divisions are four and the number of
layers nine. The key characteristics of cluster two are that it contains several units, is
narrow horizontally and medium sized vertically. The conclusions drawn from this are
that this cluster is reminiscent of the divisional structure described by Mintzberg
(1979).
Cluster three contains the largest number of numerical units at thirty eight with the
most divisions (eight) and again the most layers (ten). Ports within this group have
organisational structures with copious units, are wide horizontally and long vertically.
The seventh column contains the results of the application of Data Envelopment
Analysis applied to the inputs and outputs of the ports as discuss earlier and shown in
table 1. Column eight represents the ownership structure of the port. The ownership
structure has been divided into the three distinct types of port that exist, i.e. public,
private and joint public/private (p/p). As can be seen from columns seven and eight no
one particular cluster has a plethora of any particular ownership pattern or efficiency
rating. Thus shows that there is a fair distribution of the different types of ownership
structure and efficiency of ports among the clusters. The value of adopting cluster
analysis is that it allows for scientific analysis to establish which organisational charts

10
are similar to others without any bias from the author entering the calculation. Cluster
analysis allows for a grouping together of ports that are not dissimilar in characteristics
based solely upon the parameters specified, which in this case are units, divisions and
layers.
7. OWNERSHIP AND EFFICIENCY
Table 3 also shows the average number of ports owned for the three different port types
as defined by Cass (1996) and the average efficiency rating for each cluster. Thus, we
can see from cluster one that out of the sample of thirty one ports, nine are
public/private, seven public and four private. The ports falling into cluster one have an
average efficiency of 43.54. Of the eighteen ports that fall within cluster one, eleven
have some form of private ownership. In cluster two there are nine ports, of which eight
contain some form of private ownership, with an average efficiency of 35.84%. In
cluster three, two of the four ports contain private ownership with an average efficiency
rating of 39.92%.
Applying discriminate analysis the results were exactly the same as for cluster analysis.
This therefore confirms the results of cluster analysis as being reliable. By placing the
results from the efficiency analysis and the numerical values assigned to the
organisational charts into cluster analysis it has shown that the most efficient
organisational structure is the simple structure. The ownership structure on the other
hand does not seem to have a bearing upon the efficiency as there is no ownership type
dominant within any one cluster. The difference between the efficiency of the three
clusters is only marginal and if it is assumed that there are only two clusters the average
efficiency is 37.09%. However assuming that there are three natural clusters as
identified by cluster analysis, the implications for reform within the port industry would
show that bureaucracies do work. If a port is contemplating changing from a machine
bureaucracy to a divisional structure then the benefits in terms of efficiency are
debatable.
8. CONCLUSION
The results show that cluster analysis can be applied to analysing organisational charts.
The results enable clearly defined types of organisational structures to be identified and
also confirm that ports can be defined into the categories identified by Mintzberg
(1979) i.e. simple, divisional and machine bureaucracy. Furthermore the use of DEA as
a means of testing container port efficiency has also proven successful in helping to
highlight the characteristics of an efficient port. The results show that the simple
structure is the most efficient form of organisational structure, whilst ownership
structure does not appear to have any significant influences upon efficiency.
Predictions can also be made on the performance of a port by examining its
organisational chart to determine which of the Mintzberg (1979) categories is
applicable and the likely efficiency rating. Organisational restructuring of an inefficient
port must not been seen in its own right to be the panacea, but must go hand-in-hand
with new financing and investment. The recommendation for port managers is to
implement a structure that is simple in nature and therefore robust and reflective to
changing environments.

11
REFERENCES
(a) Books
Aaker, D. A., Kumar, V. and Day, G., S., (1995), Marketing Research, fifth edition,
John Willey & Sons, Inc.,
Abdel-Fattah, N, Gray R., Cullinane S. (1999) Road freight and privatisation: the
case of Egypt, Ashgate, Aldershot
Cass, S. (1996) Port Privatisation, Cargo Systems, IIR Publications Ltd., London
Caves, D.W., Christensen. L.R., Swanson, J.A. and Tretheway, M.W. (1982)
Economic Performance of U.S. and Canadian Railroads, In Managing Public
Enterprises, edited by Stanbury, W.T. and Thompson, F., Praeger, New York
Chapman, C. (1990) Selling The Family Silver, Hutchinson Business Books Limited,
London
Charnes, A., Cooper, W.W., Lewin, A.Y. and Seiford, L.M. (1995) Data Envelopment
Analysis: Theory, Methodology and Application, Kluwer Academic Publishers,
London
De Alessi, L. (1980) The Economic of Property Rights: A Review of the Evidence, In
Research in Law and Economics, Vol. 2, edited by Zerbe, R.O., JAL, Greenwich,
Conn., pp 1-47
Drucker, P. (1969) The age of Discontinuity: Guidelines to Our Changing Society,
Heinemann, London
Frech H.E. (1980) Property Rights, the Theory of the Firm, and Competitive Markets
for Top Decision-Makers, In Research in Law and Economics, Vol. 2, edited by
Zerbe, R.O., JAL, Greenwich, Conn., pp49-63
George, D. and Mallery. P. (1999) SPSS for Windows Step by Step, Allyn and Bacon,
London
Hair, J.F., Anderson, R.E. and Tatham, R.L and Black, W.C. (1995) Multivariate Data
Analysis with Readings, 4th ed, Prentice-Hall International (UK) Ltd, London
Mintzberg, H. (1979) The structuring of organizations, Prentice-Hall Inc.,
Englewood Cliffs, New Jersey
Szczepura, A. Davis, C., Fletcher, J. and Bousoffiane, A (1992) Applied Data
Envelopment Analysis in Health Care; the Relative Efficiency of NHS General
Practices, Warwick Business School Research Bureau, Coventry

12
(b) Journal papers
Boardman, A.E. and Vining, A.R. (1989) Ownership and performance in Competitive
Environments: A comparison of the performance of Private, Mixed and State-owned
Enterprises, Journal of Law and Economics, Vol. 32, pp 1-33
Bos, D. and Nett, L. (1991) Employee share Ownership and Privatisation: A comment,
The Economic Journal, Vol. 101, pp 966-969
Charnes, A., Cooper, WW and Rhodes, E (1978) Measuring the Efficiency of Decision
Making Units, European Journal of Operational Research, Vol.2, No.6, pp 429-444
De La Cruz, S.F. (1999) A DEA Approach to the Airport production function,
International Journal of Transport Economics, Vol. xxvi, No.2 June
De-Neufville, R and Tsunokawa, K. (1981) Productivity and returns to scale of
container port, Maritime Policy and Management, Vol. 8, No. 2, pp121-129
Everett, S. and Robinson, R. (1998) Port reform in Australia: issues in the ownership
debate, Maritime Policy and Management, Vol. 25, No. 1, pp41-62
Farrell MJ (1957) The measurement of productive efficiency, Journal of Royal
Statistical Society, Vol. 120 pp253-281
Frankel, E.G. (1991) Port Performance and Productivity Measurement, Port and
Harbours, Vol. 36, No.8, pp 11-13
Gillen, D. and Lall, A. (1997) Developing Measures of Airport Productivity and
performance: An Application of Data Envelopment Analysis, Transportation
Research -E, Vol.33, No.4, pp261-273
Heikkila E.J. (1990) Structuring a National System of Ports, Portus ete, Vol. 5 No. 3
pp19-23
Kim, M and Sachish, A. (1986) The Structure of Production, Technical Change and
Productivity in Port, Journal of Industrial Economics, Vol. 35, No. 2, pp 209-223
Martinez-Budria, E., Diaz-Armas, R., Navarro-Ibanez, M. and Ravelo-Mesa, T. (1999)
A study of the Efficiency of Spanish port authorities using Data Envelopment Analysis,
International Journal of Transport Economics, Vol. XXVI, No. 2, June
Port Development International, Public Participation, March 1999
Roe, M.S. (1999) The commercialization of East European liner shipping: the
experience of Poland, Maritime Policy & Management, Vol. 26, No.1
Roll, Y. and Hayuth, Y. (1993) Port performance comparison applying data
envelopment analysis (DEA), Maritime Policy and Management, Vol. 20, No. 2, pp
153-161

13
Sachis, A. (1996) Productivity functions as a managerial tool in Israeli ports, Maritime
Policy and Management, Vol. 23, No. 4, 341-369
Tandon, P. (1997) Efficiency of Privatised Firms - Evidence and implications,
Economic and Political Weekly, Vol. 32, No. 50, pp 3199-3212
Tongzon, J. (2001) Efficiency measurement of selected Australian and other
international ports using data envelopment analysis, Transportation Research Part A:
Policy and Practice 35 (2) (2001) pp. 113-128.
(c) Papers presented to conferences
Van Niekerk, H.C, (2000) The Restructuring Of A National Ports Network To Support
The Globalisation Of The Economy, World Confederation of Productivity Science
School of Industrial Engineering, Universidad Del Mar,
http://www.catriona.napier.ac.uk/resource/wpc10th/Niekerk.HTM 23/4/2000
(d) Other documents
De Monie, G (1987), Measuring and Evaluating Port Performance and Productivity,
UNCTAD Monographs on Port Management, Geneva
Washington Times, (2000) Global Assets Go into Private Hands at a Lively Pace, 20th
July

14
Table 1. Raw data before being analysed using DEA
Port
No. of
Containers
Total tons
throughput
Total Length of
berth
(metres)
Container berth
length
(metres)
Antwerp
3,265,000
119,788,549
62,052
8,859
Ashdod
364,000
16,194,000
3,496
480
Auckland
526,300
4,200,000
6,046
610
Felixstowe
2,461,823
30,025,285
3,972
2,773
Genoa
1,265,593
46,681,644
18,375
1,720
Halifax
435,425
14,018,831
5,948
981
Hamburg
3,546,940
75,821,000
25,941
11,615
Hawaii
1,082,346
6,567,637
9,202
3,808
Hong Kong
14,582,000
167,170,000
13,801
6,059
Houston
968,169
169,100,000
24,648
4,378
Kaohsiung
6,271,053
98,203,000
25,800
7,790
Keelung
1,621,222
26,601,000
7,730
3,090
Laem Chabang
1,425,000
13,743,000
3,350
1,600
Las Palmas
490,577
7,912,000
9,900
1,744
Le Havre
1,320,000
66,407,000
15,037
5,225
Lisbon
364,320
11,251,000
16,000
1,530
Long Beach
4,100,000
60,800,000
18,182
6,001
Los Angles
3,378,218
82,126,624
13,758
6,005
Manila
1,856,372
30,868,000
9,138
1,300
Melbourne
1,125,748
42,108,000
12,969
2,368
Mumbai
503,310
30,970,000
8,629
1,056
Port Everglades
704,390
21,000,000
7,642
2,112
Santos
859,500
39,940,386
13,004
510
Seattle
1,544,000
13,000,000
22,912
5,378
Singapore
15,100,000
313,322,000
25,884
5,810
Southampton
846,257
35,000,000
10,053
1,350
Sydney
801,081
16,450,000
5,888
950
Tacoma
1,156,500
8,612,765
11,825
2,268
Tanjung Priok
1,609,340
30,903,000
9,205
2,440
Vancouver
800,000
71,933,000
11,243
3,030
Zeebrugge
776,357
33,283,000
10,770
3,085
N.B. The column labelled ‘No. of Containers’ represent data derived from 1998 and published by Cargo
System Journal in July 1999. The source for the data represented in the remaining columns is derived
from Fairplay World Ports Encyclopaedia for the year 1998.

15
Table 2. Relative efficiency rating of sample ports
Port
Relative
Efficiency
World Ranking
(1998)
Efficiency
Ranking
Hong Kong
100
2
1
Santos
100
54
2
Singapore
100
1
3
Houston
69.86
48
4
Felixstowe
62.41
14
5
Ashdod
58.98
95
6
Manila
54.94
19
7
Vancouver
52.85
59
8
Mumbai
50.36
74
9
Los Angeles
49.28
8
10
Southampton
45.17
55
11
Genoa
44.39
34
12
Laem Chabang
40.26
29
13
Le Havre
36.47
30
14
Auckland
33.20
75
15
Sydney
32.45
58
16
Melbourne
32.27
43
17
Kaohsiung
32.23
3
18
Keelung
28.41
24
19
Tanjung Priok
27.73
25
20
Long Beach
27.63
6
21
Halifax
25.61
86
22
Zeebrugge
25.52
61
23
Hamburg
24.13
7
24
Antwerp
23.77
9
25
Port Everglades
22.70
64
26
Tacoma
19.62
37
27
Lisbon
12.08
94
28
Hawaii
11.75
41
29
Seattle
11.12
27
30
Las Palmas
10.82
76
31

16
Table 3. An analysis of ports by organisational and ownership structures using cluster and data envelopment analysis
Port
3 Clusters
2 Clusters
Units
Divisions
Layers
DEA
Ownership
Antwerp
1
1
10
3
3
23.77
p/p
Ashdod
1
1
14
9
8
58.98
public
Cluster 1
Characteristics
Felixstowe
1
1
12
10
4
62.41
private
Average Units
14
Few Units
Hawaii
1
1
7
4
3
11.75
public
Average Divisions
6
Medium horizontally
=
Simple
Hong Kong
1
1
10
5
4
100
private
Average Layers
5
Short vertically
Laem Chabang
1
1
12
8
11
40.26
p/p
p/p
8
Lisbon
1
1
23
12
7
12.08
public
Public
7
Long Beach
1
1
16
6
6
27.63
p/p
Private
3
Melbourne
1
1
9
3
7
32.27
p/p
Total
18
Mumbai
1
1
20
8
7
50.36
public
Average Efficiency
43.54
Port Everglades
1
1
11
7
3
22.7
public
Santos
1
1
19
5
5
100
private
Cluster 2
Characteristics
Seattle
1
1
20
3
3
11.12
p/p
Average Units
34
Several Units
Singapore
1
1
10
7
4
100
public
Average Divisions
4
Narrow horizontally
=
Divisional
Sydney
1
1
10
5
4
32.45
p/p
Average Layers
9
Medium vertically
Tacoma
1
1
11
5
5
19.62
public
p/p
6
Vancouver
1
1
16
5
3
52.85
p/p
Public
2
Zeebrugge
1
1
16
8
3
25.52
p/p
Private
1
Auckland
2
2
30
5
9
33.2
p/p
Total
9
Genoa
2
2
33
3
5
44.39
p/p
Average Efficiency
35.84
Hamburg
2
2
29
3
11
24.13
p/p
Kaohsiung
2
2
33
4
5
32.23
public
Cluster 3
Characteristics
Keelung
2
2
37
3
7
28.41
public
Average Units
38
Copious Units
Las Palmas
2
2
29
2
16
10.82
p/p
Average Divisions
8
Wide horizontally
=
Machine
Los Angeles
2
2
31
5
13
49.28
p/p
Average Layers
10
Long vertically
Bureaucracy
Manila
2
2
26
3
11
54.94
p/p
p/p
2
Southampton
2
2
29
5
3
45.17
private
Public
2
Halifax
3
2
42
7
11
25.61
p/p
Private
-
Houston
3
2
36
6
11
69.86
p/p
Total
4
Le Havre
3
2
36
8
10
36.47
public
Average Efficiency
39.92
Tanjung Priok
3
2
38
10
9
27.73
public
(key: p/p
=
public/private)