Transport Policy 35 (2014) 333–340

Contents lists available at ScienceDirect

Transport Policy
journal homepage: www.elsevier.com/locate/tranpol

Economic impact of port sectors on South African economy:

An input–output analysis
Young-Tae Chang a,n, Sung-Ho Shin b, Paul Tae-Woo Lee c
a
Graduate School of Logistics, Inha University, 253 Yonghyun-dong, Nam-gu, Incheon 402-751, Republic of Korea
b
Department of Logistics and Maritime Studies, Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong
c
Department of Business Administration, Soochow University, 56, Kuei-yang Street, Section 1, Taipei 100, Taiwan

ar t ic l e i nf o a b s t r a c t

Available online 16 May 2014 The port sectors in a country play an important role in its economy. This paper presents an input–output
Keywords: analysis on how the port sectors impact a concerned economy using the South African case. Moreover,
South African economy this paper reports how a rectangular Supply and Use Table system of national accounts can be converted
Port sectors to a traditional square symmetric matrix type system. A range of models, such as demand-driven,
Input–output analysis supply-driven and price models, were derived for the estimation. From these models, the production
effect together with the forward and backward linkage effects, price change effects and employment
effects were estimated to determine the impact of port sectors. The overall forward linkage effect of the
port sector was 0.97 and the backward one was 0.48, indicating that the port sector does not appear to
use other sectors much in producing its activities whereas the port sector is used relatively more by
other industries owing to its relatively high forward linkage effect. The overall impact effect of the port
sector per unit shortage on all other products was found to be 1.1705. Therefore, one unit shortage in the
port sector would have incurred a 17% loss to the entire economy in 2002. Leontief's price model was
used for the scenario that what would occur if the price of port sector's cost was increased by various
ranges from 5%, 10% and 30% to 50% and 100%.
& 2014 Elsevier Ltd. All rights reserved.

1. Introduction
The port sectors in a country play an important role in its
economy, which is particularly true when a country develops its
port as a regional hub. For example, Singapore and Hong Kong have
developed their ports as regional hubs in Asia (Lam, 2011), and the
Netherlands has developed its ports as a regional hub in Europe.
Similarly, Los Angeles and Long Beach function as regional hubs in
North America. These ports make utmost efforts in continuously
maintaining their ports as hubs and improving them to make them
more attractive than competing ports due to their important roles in
economic development (Lam and Yap, 2011a, 2011b). Most of these
regional hub ports are located in the northern hemisphere. Ports in
the southern hemisphere have not reached the status of becoming
regional hubs due to relatively underdeveloped economic integration
in that region. In recent years, some countries in the southern
hemisphere have attempted to develop their ports as regional hubs.
On the other hand, the idea of a hub port strategy cannot be justified
n
Corresponding author.
E-mail address: ytchang@inha.ac.kr (Y.-T. Chang).
when people are unsure if this project can bring enough economic
benefits to the country. A more problematic case is that people do
not understand how port sectors contribute to their economy in
terms of the impact of port sectors on other industries, employment
effects, price change effects, etc. South Africa is the case of this kind
when they considered developing a hub port shifting from a
traditional set of gateway port system (Notteboom, 2011). The main
container ports in South Africa are Durban, Cape Town and Port
Elizabeth, whereas East London and Richards Bay handle small
container volumes. The country plans to develop a hub port in a
new site, called Ngqura, which is the home town of President Nelson
Mandela. Despite this hub development strategy in South Africa, the
government and people have wondered how important their port
sectors are in their economy in terms of the impact on other
industries, employment and price effects. When a country or
a region attempts to develop a major port industry, decision-
makers should be confident that their new investments will bring
sufficient economic impacts on the country and the region in terms
of employment effect, value-added amounts, production-inducing
effects on other industries, etc. This was the case of South Africa,
when the country considered developing a new hub port not only to
resolve chronic congestion problem in its representing container

http://dx.doi.org/10.1016/j.tranpol.2014.04.006
0967-070X/& 2014 Elsevier Ltd. All rights reserved.
334 Y.-T. Chang et al. / Transport Policy 35 (2014) 333–340
port, namely Durban, but also to stimulate its economy based on this
new hub port. Therefore, the economic impacts had to be analyzed to
address these issues.
Surprisingly, despite that there are several studies in the area of
port sectors' economic impact analysis available, the extant studies are
too narrowly focused on the impact of the sectors (Moon, 1995),
merely on employment effect of port sector in regional cases (Hughes,
1997; Musso et al., 2000; Acciaro, 2008), or are focused on broader
maritime industries than port sectors (Kwak et al., 2005) or only the
shipping sector (Van Der Linden, 2001). One study reported only a
descriptive method without a detailed methodology so that replica-
tion of the research elsewhere is infeasible (Van Der Linden, 2001).
This paper intends to contribute to the literature by presenting an
input–output analysis on how the port sectors can affect a concerned
economy using the South African case. Musso et al. (2000) proposed a
technique to assess the employment impact of the Port of Genoa
dealing with several types of ships. But this paper handles more
comprehensive economic impacts of port sector in a nation. Moreover,
this paper outlines how the rectangular Supply and Use Table system
of national accounts can be converted to a traditional square sym-
metric matrix type system, using the national account data in South
Africa. In 1993, South Africa changed their national account system to
a Supply and Use Table system recommended by the United Nations
(United Nations, 1993). Therefore, this converting approach is differ-
entiated from the previous port impact studies.
The next section explains the methodology and data. Section 3
presents the results and Section 4 discusses the implications and
concludes the paper.
2. Methodology and data
The main methodology of this paper is static input–output
analysis. There are various methods in assessing economic impact
analysis of port sector in the literature. See Acciaro (2008) and
Danielis and Gregori (2013) for detailed description of the various
methods in the literature. Two major methods in the economic
impact analysis are input–output analysis and the computable
general equilibrium (CGE) model. The former method is more
widely used in the impact analysis of port sector (see Danielis and
Gregori, 2013; Kwak et al., 2005; Lee and Yoo, 2014; Morrissey and
O'Donoghue, 2013; Van Der Linden, 2001) compared with the
latter (see Lee et al., 2011; Lee et al., 2012). DeSalvo (1994)
contends that port economic impacts are wrongly estimated as
they do not consider the price changes in local area production if
there were not any port services available. Therefore, he recom-
mends using a supply-demand model. Recent researches in using
the I-O model and CGE model are focused more on the environ-
mental estimation of sectors (Lee et al., 2013; Neuwahl et al., 2008;
Su et al., 2013; Su and Ang, 2014). As Acciaro (2008) points out, it
is true with input–output analysis that a certain level of subjective
classification of disaggregating port sectors from the general
industry classification system is needed. However, compared with
other methodologies in capturing various inter-linkages between
sectors and also production-inducing effect, value-added effect,
employment effect, etc. the input–output analysis is more proper
to be used for estimating economic impacts and contribution to
the South African economy. In addition, using the Supply and Use
table, the input–output analysis is the most proper one to capture
these effects. The Input–Output (IO) model shows the relationship
between the productive sectors of a given economy in a linear,
inter-sectoral model. The relationship between the productive
sectors and demand can be expressed as follows:
N N
X i ¼ ∑ X ij þ F i ¼ ∑ aij X j þ F i ð1Þ
j¼1 j¼1
or
N N
X j ¼ ∑ X ij þV j ¼ ∑ r ij X i þV j ð2Þ
i¼1 i¼1
where Xi is the total gross output in sector i¼1,…, N: aij are the direct
input or technical coefficients that divide Xij, the inter-industry
purchases of producing sector i from supply sector j by Xj, which
is the total gross output in sector j; rij are the direct output
coefficients that divide Xij, which are the inter-industry purchases
of producing sector i, from the supply sector j by Xi, which is the
total gross output in sector i; and Fi is the final demand for
products in sector i and Vj is the final value-added by sector j.
Therefore, Eq. (1) shows the demand-driven model as viewing IO
tables vertically, whereas Eq. (2) expresses the supply-driven
model as viewing IO tables horizontally.
Eq. (1) can be rewritten in an abbreviated matrix form as X¼
(I  A)  1F (Chiang, 1984). I denotes the N  N identity matrix and
(I  A)  1 is called the Leontief inverse matrix. The standard
demand-driven model of this matrix form, however, cannot assess
the net effects of port sector activities precisely. Hence, the
individual port sector needs to be handled as exogenous and
placed into the final demand group (Han et al., 2004; Kwak et al.,
2005). Therefore, the port sector-based IO model or exogenized IO
model for port sector can be expressed as Xe ¼(I Ae)  1
(Fe þAmXm), where subscript e refers to an exogenized matrix
and m refers to the port sector. Assuming ΔFe ¼0, results in
ΔXe ¼ ðI Ae Þ  1 AM ΔXM ð3Þ
Eq. (3) can be used to estimate the relationship of inter-industries
impacted by a change in port investments, i.e. the production
inducing effect.
Similarly, the exogenized Leontief's price model can be used for
the port price change effect and the exogenized supply driven
model can be used for the impact of limited capacity. The models
can be explained as follows:
 4 4
Δ P ¼ ðI Ae' Þ  1 A M Δ P M ð4Þ
 4
where, Δ P is the matrix of normalized price, and A M is the port
sector's matrix treated as exogenous.
The equation shows that the port sector can be treated as
exogenous and placed into the primary input group. This is
a rewritten form of the conventional Leontief price model without
price changes in the value-added sector. If it is assumed that
the cost change of each sector can be transferred completely and
the annual production of each sector is given, one can assess the
effects of a change in wholesale price on the economic system
caused by a cost change in the port sector using the following
equation (Kwak et al., 2005):
ΔXe' ¼ R M ΔXM ðI  Re Þ  1 ð5Þ
1
where R is the output coefficient matrix and (I  Re) is the
output inverse matrix of which elements ij¼∂Xj/∂Vi represent the
total direct and indirect requirements in sector j per unit of final
value added in sector i (Han et al., 2004; Kwak et al., 2005). The
port sector is also treated as exogenous to disaggregate its impact
on other industries. This equation can enable an estimation of the
impacts of a unit shortage in the port sector on the output of all
other sectors, and can be used as a basis to estimate the macro-
economic impact of the limited capacity.
Up to 1993, South Africa published traditional IO tables. Since 1993,
the Supply and Use Tables (SUTs) have been used according to the
recommendations of UN 1993 System of National Accounts (SNAs)
(United Nations, 1993). The 2002 SUT was used in the present study,
as these tables are the most updated and detailed data published by
Stat SA during the timing of this research (Statistics South Africa,
2006). The SUT shows how products have been supplied and used by
 Y.-T. Chang et al. / Transport Policy 35 (2014) 333–340 335

industry in the benchmark year. Unlike the squared characteristics of For example
the former IO tables, the SUT shows different number of rows and 0 1
g1 0
columns in the respective Supply and Use Tables. The Supply table 4 B C
gn ¼ @ ⋱ A ð6Þ
shows 153 products in the rows and 94 industries in the columns,
whereas the Use Table shows 95 products in the rows and 94 0 gn
industries in the columns in 2002. It is important to convert the As explained in the notations, the use coefficient matrix and the
SUT into square symmetric IO tables to analyze the macroeconomic market share matrix are first defined:
impact arising from port industry's economic activities.
4 1
To carry out this task, it is vital to adjust the different number of B ¼ U U gn ð7Þ
rows or columns between the Supply and Use Tables. For example, as
is often the case, when analyzing the inter-industry economic impact, D ¼ M0 U q
b 1
ð8Þ
it is important to convert the SUT into the symmetric industry IO
tables. This process estimates how Supply products along the same First, it is important to equalize the dimension of the rows
number of industries have been used in the Use Table for these (number of products: 153 in year 2002) in the Supply Table with
industries. This estimation can be undertaken either by collecting the dimension of the rows (number of products: 95 in year 2002)
further detailed data in addition to SNA data or by mathematically in the Use Table. Once the same number of products or rows is
calculating the conversion. Since Stat SA was in its early stages when made in both the Supply Table (henceforth, called the adjusted
publishing the SUT, it did not appear to possess sufficient levels of Supply Table, and denoted as M referring to the adjusted make
information and data enough to produce IO tables. Therefore, in the matrix of the Supply Table) and Use Table, the matrix multi-
present study, the mathematical conversion was selected due to the plication of Eq. (7) can be calculated by multiplying the U matrix
lack of detailed data. with the industry output in the Use Table:
Basically, two methods can be used to combine the use and 2 3
s1 1 s1 2 … s1 94
supply matrices mathematically to generate the traditional symmetric 6 s2 1 s2 2 … s2 94 7
6 7
input–output matrix. These methods are based on either an industry S ¼ 6 7 ð9Þ
ð15394Þ 4 ⋮ ⋮ ⋱ ⋮ 5
technology assumption or a commodity technology assumption. The
industry technology assumption assumes that the inputs are con- s153 1 s153 2 … s153 94
sumed in the same proportion by every product produced by a given 2 3
industry. This assumption has been used in many countries based on u1 1 u1 2 ⋯ u1 94
6 u2 1 u2 2 ⋯ u2 94 7
the recommendations made by the 1968 SNA, mainly for two 6 7
U ¼6 7 ð10Þ
attractive reasons. First, the method always generates positive sym- ð9594Þ 4 ⋮ ⋮ ⋱ ⋮ 5
metric input–output tables. Second, it is also applicable to the case of u95 1 u95 2 ⋯ u95 94
rectangular input–output tables. In contrast, the commodity technol-
20 13
ogy assumption assumes that the input structure of the technology m1 1 … m1 94
that produces a given product is the same regardless of where it is 6B ⋮ ⋱ ⋮ C7
M ¼ 4@ A5 ð11Þ
produced. Although this assumption appears economically more ð9594Þ
m95 1 ⋯ m95 94
reasonable than the industry technology assumption, it is not used
widely because it tends to generate negative symmetric input–output 20 13
m1 1 … m1 95
tables and requires the make and intermediate matrices of the Use 0 6B C7
M ¼ 4@ ⋮ ⋱ ⋮ A5 ð12Þ
Table to be squared (United Nations, 1999). In other words, the ð9495Þ
commodity technology assumption requires that the domestic pro- m94 1 ⋯ m94 95
duction part of the supply table (herein refer to as matrix, M) should Eq. (12) shows the transposed matrix of M to allow the multi-
be invertible to calculate the input–output table. M is invertible only if plication of the Use Table and Supply Table and produce a product-
M is square or the number of industries must equal the number of by-product input–output table. The product-by-product input–
products. Since the Use Table in South Africa is rectangular and not output table can be calculated using Eqs. (7) and (8), Eq. (10)
squared due to the different number of rows and columns between and Eq. (12) as follows:
the product and industry numbers, respectively, the industry technol- 2 3
ogy assumption was used to calculate the input–output tables from a1 1 a1 2 … a1 95
6 a2 1 a2 2 … a2 95 7
the SU tables. 6 7
A ¼ B d D ¼6 7 ð13Þ
The mathematical process adopted can be explained using the ð9595Þ ð9594Þ ð9495Þ 4 ⋮ ⋮ ⋱ ⋮ 5
same notations from the United Nations (United Nations, 1999): a95 1 a95 2 … a95 95
Notations
When completing the product-by-product IO table, the final
m number of products demand can be calculated using the following equation.
n number of industries YC ¼ ðI  BDÞ U q
Um  n intermediate matrix of the Use Table (product by industry)
where; YC is final demand of products ð14Þ
Bm  n use coefficient matrix (product by industry)
Mm  n make matrix (product by industry), part of the supply Matrix S is the rows and columns in the Supply Table. As can be
matrix describing domestic production seen in the dimensions of the matrix, the number of products in
Dsubn  m market share matrix (industry by product) the rows in 2002 was 153 and the number of industries in the
gn vector of industry output columns was 94. Similarly, the number of products in the Use
qm vector of product output Table was 95 and the number of the industries in matrix U was 94.
_
g diagonal matrix of industry output To make possible the conversion of the SUT into a symmetric IO
_
q diagonal matrix of product output table, which was target in the present study, the number of rows
in the Supply Table (153) was first equalized with the number of rows
The matrix with a ‘\widehat’ is a diagonal version of the vector in the Use Table (95) by reducing the number in the Supply Table to
with the same notation; all off-diagonal elements are zeros. the same number in the Use Table. This is shown in matrix M. Once
336 Y.-T. Chang et al. / Transport Policy 35 (2014) 333–340
the same number of rows in the new M matrix is equalized with the
number of rows in the Use Table, the matrix multiplication can be
used to derive the symmetric IO table. First, the M matrix is
transposed before multiplying it with the Use Table matrix, U, as
shown in Eq. (12). This transposed matrix, M', is now composed of
industry rows and product columns. B (the use coefficient matrix) is
multiplied by D (the market share matrix) to calculate the inter-
mediate product-by-product matrix, A, as shown in Eq. (13).This
intermediate product-by-product matrix, A, is the most fundamental
one to make calculations for a macroeconomic impact. Once the A
matrix is calculated, the next process can be performed through an
exogenous treatment of the port sector and follow the price model
and supply-driven model explained in Eqs. (4)–(5).
After exogenizing the port sector, the general procedures of
calculating the technology matrix, (I  Ae) and its inverse matrix,
(I  Ae)  1 can then be performed (Chiang, 1984). By multiplying
the inverse matrix, (I  Ae)  1 with the exogenized port sector
matrix, Am, the employment effect, income effects, linkage effects,
production inducing effects, etc. can be estimated once the
respective data can be collected across the product rows in
employment, income, production, etc. of the SUT. The employment
effect, for example, can be denoted in the following notation:
L ¼ ^lðI  Ae Þ  1 Xm þ ^l U Xm ð15Þ
where, L is the employment effect induced by the port sector's
activities; ^l the diagonal matrix of product groupwise employment
coefficient; ðI  Ae Þ  1 the inversed technology matrix after exo-
genizing the port sector, Xm the exogenized port sector vector.
3. Results
To derive the adjusted make matrix, M, the original Supply
product classification was regrouped based on a similar classifica-
tion to the Use Table. Eqs. (7) and (8), Eqs. (12) and (13) were
used to calculate the symmetric Product-by-Product IO matrix.
From this matrix, the inter-industry linkage effect, such as the
forward linkage and backward linkage effects, could be derived
and estimated, and demand and supply driven model could be
derived to estimate the production-inducing effects and supply
shortage effects, respectively. Finally the Leontief price model
could be used to estimate the impact of a change in the port
sector's price on the other products and employment-inducing
effects etc.
The conversion from SUT to symmetric IO tables is the first
critical task to estimate the macroeconomic impact. Before placing
a mathematical application into the STAT SA's SUT, an attempt
was made to validate if the mathematical calculation process
was correct using the Irish SUT and their converted IO
tables (Government of Ireland, 2004). The calculation process
was found to be correct. Once the SUT is converted into symmetric
Input–Output tables, the next important task in estimating the
macroeconomic impact is to disaggregate the port sector's eco-
nomic activities from the general Standard Industrial Classification
(SIC) product groups because the port sector's activities are
scattered in various sectors of the product groups due to the SIC's
broad classification system. This is a general problem faced by
economists estimating the macroeconomic impact of a specific
industry. To disaggregate the port sector, this study examined how
the SIC is classified further into smaller group products by STAT SA
and came to the conclusion that the port sector is dispersed into
five major product groups, such as other transport (I77), buildings
(I83), other construction (I84), trade (I85) and transport services
(I87). Of these, how each group consists of its respective subgroup
products and how much the economic activities were incurred
was traced by examining further details of the other statistics data
published by STAT SA and other relevant government agencies. As
there are a range of sources available over these times, it was
assumed that the proportion of port sector's activity in each SIC
group would have been stable. The proportion was estimated after
breaking down these SIC group products and calculating the
proportion of the port sector in each SIC group. Based on the
disaggregation process, the coefficient matrix, Ae, was calculated
after exogenizing the port sector.
The production-inducing effect was calculated and the result is
presented in Tables 1 and 2. For example, if the port sector is
produced by a single unit, the insurance sector would have 10% of
impact by the production in port sector. Table 2 lists top 10
production-inducing effect groups based on the 2002 IO table.
Insurance services, petroleum products, other business services,
and communications are in descending order of the most
production-inducing groups. The insurance sector would have
10% impact by a one unit product (i.e. one million Rand) in the
port sector. Table 2 also shows the direct, indirect and induced
production-inducing amounts according to product group caused
by the port sector. The overall direct production-inducing effect by
the port sector in 2002 was estimated to be 4.026 billion Rand and
the indirect effect was 3.854 billion Rand, giving an induced effect
(direct þ indirect) of 7.88 billion Rand. The multiplier effect was
estimated to be 1.95710. This means that if the port sector spends
one unit of money, e.g. one million Rand, this expenditure would
induce a 1.95710 units, i.e. 1,957,170 Rand production for the entire
SA economy.
Table 3 lists the backward and forward linkage effects by
product group. The overall forward linkage effect of the port
sector is 0.97 and the backward one is 0.48 indicating that the
port sector does not appear to use other sectors much in produ-
cing its activities due to the low backward linkage effect, whereas
the port sector is used relatively more by other industries owing to
its relatively high forward linkage effect. The top 10 forward
linkage effect product groups are as follows: other business
services; transport services; iron and steel products; other mining
products; insurance services; etc. The top 10 backward linkage
effect groups are trade services; handbags; containers of paper;
buildings; rubber tyres; etc.
The supply-driven model can be used to estimate how a supply
restriction, e.g. not developing the port sector affects the other
industries' activities. This is again to exogenize the port sector, as
was the case with the demand-driven model, and evaluate the
impact of a shortage in the port sector on the output of all other
sectors. This model can be used as a basis to define the shortage or
failure costs of port production. As in the supply-driven model
equation, a shortage in the port sector can be estimated using
2002 data. The result is listed in Table 4. The table shows how
a unit shortage in the port sector can impact other product groups
from the coefficient figure. In addition, the table shows the direct
and indirect impact amount for the shortage of the port sector by
product group. If there were no port sector activities, it would have
resulted in a 1.481 billion Rand direct loss to the entire economy
and 1.734 billion Rand indirect loss, totaling 3.215 billion Rand.
Again, the overall impact effect per unit shortage of port sector on
all other products is 1.1705. Therefore, one unit shortage in the
port sector would have incurred 17% of loss to the entire economy
in 2002. Without port producing activities in 2002 (4.3 billion
Rand), the entire economy of SA would have suffered 3.22
billion Rand.
Finally, the employment effect induced by port investment was
calculated. The results are listed in Table 5. First, this study
calculated how many people are employed per unit amount
(million Rand) of each product group's economic activity. The
employment coefficient could be identified in a similar manner as
the 58 product groups based mainly on two sources published by
 Y.-T. Chang et al. / Transport Policy 35 (2014) 333–340 337

Table 1
Production-inducing effect.

Code Product Coefficient Code Product Coefficient Code Product Coefficient Code Product Coefficient
No. no. no. no.

P1 Agricultural products 0.00628 P25 Footwear 0.00091 P49 Iron and steel 0.02552 P73 Radio and 0.00803
products television products
P2 Coal and lignite 0.01296 P26 Wood products 0.0075 P50 Non-ferrous metals 0.00982 P74 Optical 0.00369
products instruments
P3 Gold and uranium 0 P27 Paper products 0.00984 P51 Structural metal 0.0096 P75 Motor vehicles 0.00518
ore products products
P4 Other mining 0.03720 P28 Containers of paper 0.0078 P52 Treated metal 0.00278 P76 Motor vehicles 0.03836
products products parts
P5 Meat products 0.00095 P29 Other paper 0.00365 P53 General hardware 0.00105 P77 Other transport 0.02982
products products products
P6 Fish products 0.00009 P30 Published and 0.01009 P54 Other fabricated 0.00787 P78 Furniture 0.00295
printed products metal products
P7 Fruit and vegetables 0.00009 P31 Recorded media 0.00044 P55 Engines 0.00161 P79 Jewelry 0.00001
products products
P8 Oils and fats 0.00047 P32 Petroleum products 0.08437 P56 Pumps 0.00115 P80 Other 0.02385
products manufacturing
P9 Dairy products 0.00055 P33 Basic chemical 0.01084 P57 Gears 0.00244 P81 Electricity 0.01713
products
P10 Grain mill products 0.00036 P34 Fertilizers 0.00163 P58 Lifting equipment 0.00248 P82 Water 0.00293
P11 Animal feeds 0.00085 P35 Primary plastic 0.01085 P59 General machinery 0.00152 P83 Buildings 0.01813
products
P12 Bakery products 0.00008 P36 Pesticides 0.00064 P60 Agricultural 0.00022 P84 Other 0.01628
machinery constructions
P13 Sugar products 0.00029 P37 Paints 0.00842 P61 Machine-tools 0.00057 P85 Trade services 0.02977
P14 Confectionary 0.00003 P38 Pharmaceutical 0.0017 P62 Mining machinery 0.00427 P86 Accommodation 0.00816
products products
P15 Other food products 0.00056 P39 Soap products 0.00461 P63 Food machinery 0.00033 P87 Transport services 0.04248
P16 Beverages and 0.00120 P40 Other chemical 0.00634 P64 Other special 0.00331 P88 Communications 0.06010
tobacco products products machinery
P17 Textile products 0.00279 P41 Rubber tires 0.01241 P65 Household 0.00016 P89 FSIM 0.01096
appliances
P18 Made-up textile 0.00401 P42 Other rubber 0.00159 P66 Office machinery 0.00027 P90 Insurance services 0.10029
products products
P19 Carpets 0.00026 P43 Plastic products 0.01159 P67 Electric motors 0.00330 P91 Real estate 0.03865
services
P20 Other textile 0.00033 P44 Glass products 0.00139 P68 Electricity apparatus 0.00418 P92 Other business 0.07764
products services
P21 Knitting mill 0.00031 P45 Ceramic ware 0.00154 P69 Insulated wire and 0.00424 P93 General 0.00152
products cable Government
services
P22 Wearing apparel 0.00635 P46 Ceramic products 0.00515 P70 Accumulators 0.00258 P94 Health and social 0.00219
work
P23 Leather products 0.00055 P47 Cement 0.00654 P71 Lighting equipment 0.00152 P95 Other services/ 0.02526
activities
P24 Handbags 0.00002 P48 Other non-metallic 0.01253 P72 Other electrical 0.00405 Total 0.95717
products products

Table 2
Top 10 production-inducing effects.

Rank Code no. Products Production-inducing effects by port sector

Production-inducing coefficient Indirect impact Direct impact Total (directþindirect)

1 P90 Insurance services 0.10029 403.8 215.6 619.4

2 P32 Petroleum products 0.08437 339.7 260 599.7
3 P92 Other business services 0.07764 312.6 157.5 470.1
4 P88 Communications 0.06010 242 78.5 320.5
5 P87 Transport services 0.04248 171.1 38.5 209.6
6 P91 Real estate services 0.03865 155.6 45.1 200.7
7 P76 Motor vehicles parts 0.03836 154.4 113 267.4
8 P4 Other mining products 0.03720 149.8 17.3 167.1
9 P77 Other transport products 0.02982 120.1 97.2 217.3
10 P85 Trade services 0.02977 119.9 65.2 185.1

Total 0.95170 3854 4026 7880

(Unit: million Rand).

Statistics South Africa (Statistics South Africa, 2003, 2008). Com- group coefficients, it was important to recalculate the symmetric
bining these sources, 58 product groups were identified as the IO matrix from the original 95 groups to 58 groups to be consistent
most detailed employment coefficient. Based on these 58 product with the labor product groups that had been identified. From the
338 Y.-T. Chang et al. / Transport Policy 35 (2014) 333–340

Table 3
Top 10 forward linkage effect and backward linkage effect.

Rank Code no. Products Forward linkage effects Rank Code no. Products Backward linkage effects

1 P92 Other business services 3.71830 1 P85 Trade services 4.22291

2 P87 Transport services 3.02732 2 P24 Handbags 2.26039
3 P49 Iron and steel products 3.00927 3 P28 Containers of paper 1.55900
4 P4 Other mining products 2.98499 4 P83 Buildings 1.39622
5 P90 Insurance services 2.74251 5 P41 Rubber tyres 1.32985
6 P1 Agricultural products 2.49316 6 P51 Structural metal products 1.30243
7 P88 Communications 2.49025 7 P40 Other chemical products 1.29466
8 P91 Real estate services 2.26539 8 P11 Animal feeds 1.24911
9 P32 Petroleum products 2.10116 9 P49 Iron and steel products 1.20461
10 P35 Primary plastic products 2.07158 10 P52 Treated metal products 1.19397

Total (95 sectors) 0.97003 Total (95 sectors) 0.47623

Table 4 the macro-economy of SA. Various types of models, such as the

Top 10 supply shortage effects.(Unit: million Rand). demand-driven model, the supply-driven model and the price
Rank Code Product Supply Direct Indirect Total
model, were used to address a range of issues. From the findings of
no. shortage impact impact economy- the models, the following implications can be drawn to assist the
coefficient amount amount wide policy-formulation in a future port investment for the South
impact African economy.
The overall forward linkage effect of the port sector was 0.97
1 P85 Trade services 0.1310 177.98 194.03 372.02
2 P4 Other mining 0.1160 283.84 171.87 455.71 and the backward one was 0.48, indicating that the port sector
products does not appear to use other sectors much in producing its
3 P49 Iron and steel 0.0700 81.76 103.65 185.4 activities due to the low backward linkage effect, whereas the
products port sector is used relatively more by other industries owing to its
4 P83 Buildings 0.0603 84.73 89.34 174.07
5 P92 Other business 0.0599 79.5 88.67 168.16
relatively high forward linkage effect. Backward linkage effects are
services strongly induced by industries with high intermediate input
6 P87 Transport 0.0557 45.76 82.58 128.34 coefficients, for instance, manufacturing industries. On the other
services hand, forward linkages are strongly induced generally by primary
7 P88 Communications 0.0520 61.87 76.98 138.85
and material industries, whose outputs are used by other indus-
8 P2 Coal and lignite 0.0493 117.66 73.04 190.7
products tries as intermediate goods (Kwak et al., 2005). In this respect, the
9 P32 Petroleum 0.0487 51.97 72.21 124.18 magnitude of 0.97 in the forward linkage effect of South Africa is
products noteworthy compared with other countries which are strong in
10 P93 General 0.0457 64.21 67.63 131.84 maritime power, for instance, Ireland of having 0.62 (Morrissey
Government
services
and O'Donoghue, 2013). This implies that South Africa has a strong
potential to develop its new hub port in view of its relatively high
Total group (95 groups) 1.1705 1481 1734 3215
forward linkage effect as the development of the new port can be
more actively utilized by other relevant industries as intermediate
goods and services. The overall impact effect of the port sector per
unit shortage on all other products was found to be 1.1705.
Therefore, one unit shortage in the port sector would have
employee coefficient per unit supply by product group, the incurred a 17% loss to the entire economy in 2002. This also
_
product-by-product employment-inducing effect (l  (I  A)  1) indicates that the economic significance of port sector in South
_
was calculated, where l refers to the diagonal matrix with the Africa is as high as the one in Korea in terms of the shortage effect
diagonal elements being the 58 product groupwise employment (Kwak et al., 2005). Leontief's price model was used for the
coefficient. Following this process, the employment-inducing scenario that what would occur if the price of port sector's cost
effect per port supply of million Rand was calculated, as listed in was increased by various ranges from 5%, 10% and 30% to 50% and
Table 5. The table shows the direct and indirect employment effect 100%. The top 10 impacted product groups by the cost change are
by the 58 product groups both in terms of the absolute number of transport services; other constructions; trade services; coal and
people per group and per million Rand of port supply in 2002. lignite products; and sugar; cement; other mining products;
From the table, it can be seen that 14,209 people were employed buildings; iron and steel products; other non-metalic products.
directly by the port sector in 2002 and 36,418 people were In the case of a 5% change in cost, their impacts on the top five
employed indirectly, making the total of 50,627 people employed product groups would be 38%, 33.1%, 25.5%, 8.1%, and 6.5%. Finally,
by the port sector. The per million Rand employment effect was the employment effect induced by port investment was calculated.
12.57 in 2002 including the direct and indirect effects. The results showed that 14,209 and 36,418 people were employed
directly and indirectly, respectively, by the port sector in 2002,
making a total of 50,627 people employed in the port sector. The
4. Implications and conclusions per million Rand employment effect was 12.57 in 2002 including
the direct and indirect effects.
This paper attempted to estimate the macroeconomic impact These estimation results may draw the following implications.
arising from the port sector using an Input–Output Analysis model. First of all, as South Africa plans to develop a new container hub
After converting the SUT tables to symmetric IO tables, the port port (the port of Ngqura) and logistics distribution center, the
sector was exogenized to estimate the port sector's contribution to newly developed container hub port can be more widely utilized
 Y.-T. Chang et al. / Transport Policy 35 (2014) 333–340 339

Table 5
Employment-inducing effects.

Product Direct Indirect Directþindirect Directþ indirect

employee employee employee employee
per million Rand

Agriculture, hunting, forestry and fishing 35 1070 1104 0.27

Mining and quarrying 169 2529 2697 0.40
Manufacturing 2076 6928 9005 2.24
Electricity, gas and water supply 151 395 546 0.14
Construction 1208 2356 3564 0.89
Wholesale-, retail- and motor trade, hotels and restaurants 6637 11,798 18,435 4.58
Transport, storage and communications 573 2767 3340 0.83
Finance, real estate and business services 3233 9069 12,302 3.06
Community, social and personal services 163 576 739 0.18

Total product groups 14,209 36,418 50,627 12.57

by other related industries as reflected in the relatively high Acknowledgments

forward linkage effect. To further intensify its inter-industry
linkage effect not only in vertical integration through forward
and backward linkage effect, but also in horizontal integration
within maritime industry, policy makers need to develop a short
sea shipping network to provide feeder service for East and West
Africa and ocean-going routes with Asian and European regions.
Current container ports in South Africa do not function as the
regional hub, therefore, there is a low level of integration with
other sectors and also within the maritime sector. To enhance
more interaction effect between sectors and also between regions
once the hub is developed, the short sea shipping network
between the hub and feeder ports is likely to play an important
role for the higher linkage effect. Furthermore, to secure integrated
inland transportation network and lessen the cargo-handling
burden on the seaboard of the country, dry ports in inland areas
can be explored. Along the same context, strategic option of
relaying services between South Africa and South America can
provide more opportunities to increase international trade
between the southern hemisphere continents. Moreover, as
Acciaro (2008) found that the employment effect generated by
the port sector is higher in regional economy than the total
national one, a dedicated container port, Ngqura in South Africa
can contribute more to enhancing the employment effect in the
regional level.
Some caveats should be taken when interpreting the results of
IO analysis. One of the limitations was to purely distill the port
sector's economic activities, which are scattered in a variety of
industries and product groups due to the broad level of the
Standard Industry Classification System of STAT SA. Despite the
research team's endeavors to identify the port sector's economic
activities, there are some uncertainties regarding the extent of
disaggregating the port sector from the current 95 product groups.
The 2002 SUT is the most detailed data source of collecting the
National Accounts, amounting to 153 products and 94 industries
for the Supply Table and 95 products and 94 industries in the Use
Table. Nevertheless, even these details of the classification are not
yet sufficient for economists to separate the port sector from all
other products and industries. Therefore, STAT SA is advised to
collect the port sector's data in more details in the future. Another
limitation was the inconsistency of the statistical data in general
between the different sources. This is particularly true with the
labor survey data from different sources. Similar to the IO table
data, employment survey data in the port sector should be
collected separately as an independent study in the future because
the general transport industry's survey data do not show the
necessary level of details for the port sector. All these limitations
remain an avenue for future research.
This work was supported by TRANSNET, South Africa. The
authors appreciate comments and support by Mr. Mathew Flynn
during this research.
References
Acciaro, M., 2008. The role of ports in the development of Mediterranean islands:
the case of Sandinia. Int. J. Transp. Econ. 35 (3), 295–323.
Chiang, A.C. Fundamental methods of mathematical economics. McGraw-Hill
International Edition; 1984.
Danielis, R., Gregori, T., 2013. An input–output based methodology to estimate the
economic role of a port: the case of the port system of the Friuli Venezia Giulia
Region, Italy. Marit. Econ. Logist. 15 (2), 222–255.
DeSalvo, J.S., 1994. Measuring the direct impacts of a port. Transp. J. (Am. Soc.
Transp. Logist. Inc) 33 (4), 33–42.
Government of Ireland. ESA95 supply, use and input-output table 2002; 2004.
Han, S.Y., Yoo, S.H., Kwak, S.J., 2004. The role of the four electric power sectors in
the Korean national economy: an input–output analysis. Energy Policy 32 (13),
1531–1543.
Hughes, W.R., 1997. A comparison of economic impacts with the use of economic
base and input–output methodologies. Environ. Plan. A 29, 673–684.
Kwak, S.J., Yoo, S.H., Chang, J.I., 2005. The role of the maritime industry in the
Korean national economy: an input–output analysis. Mar. Policy 29 (4),
371–383.
Lam, J.S.L., 2011. Patterns of maritime supply chains: slot capacity analysis. J. Transp.
Geogr. 19 (2), 366–374.
Lam, J.S.L., Yap, W.Y., 2011a. Container port competition and complementarity in
supply chain systems: evidence from the Pearl River Delta. Marit. Econ. Logist.
13 (2), 102–120.
Lam, J.S.L., Yap, W.Y., 2011b. Dynamics of liner shipping network and port
connectivity in supply chain systems: analysis on East Asia. J. Transp. Geogr.
19 (6), 1272–1281.
Lee, T.C., Lee, P.T.W., Chen, T., 2012. Economic impact analysis of port development
on the south african economy. S. Afr. J. Econ. 80 (2), 228–245.
Lee, T.C., Chang, Y., Lee, P.T.W., 2013. Economy-wide impact analysis of a carbon tax
on international container shipping. Transp. Res. Part A: Policy Pract. 58 (0),
87–102.
Lee, T.C., Wu, C., Lee, P.T.W., 2011. Impacts of the ECFA on seaborne trade volume
and policy development for shipping and port industry in Taiwan. Marit. Policy
Manag. 38 (2), 169–189.
Lee, M., Yoo, S., 2014. The role of the capture fisheries and aquaculture sectors in the
Korean national economy: an input–output analysis. Mar. Policy 44 (0),
448–456.
Moon, S.H., 1995. Port economic impact model (PIM) and its planning applications.
Marit. Policy Manag. 22 (4), 363–387.
Morrissey, K., O'Donoghue, C., 2013. The potential for an Irish maritime transporta-
tion cluster: an input–output analysis. Ocean Coast. Manag. 71 (0), 305–313.
Musso, E., Benacchio, M., Ferrari, C., 2000. Ports and employment in port cities. Int.
J. Marit. Econ. 2 (4), 283–311.
Neuwahl, F., Lö schel, A., Mongelli, I., Delgado, L., 2008. Employment impacts of EU
biofuels policy: combining bottom-up technology information and sectoral
market simulations in an input–output framework. Ecol. Econ. 68 (1–2),
447–460.
Notteboom, T., 2011. An application of multi-criteria analysis to the location of
a container hub port in South Africa. Marit. Policy Manag. 38 (1), 51–79.
Statistics South Africa, 2003. Discussion Paper Comparative Labor Statistics Survey
of Employment and Earnings in Selected Industries, December 2002.
340 Y.-T. Chang et al. / Transport Policy 35 (2014) 333–340

Statistics South Africa, 2006. Supply and Use Tables: Final Supply and Use Tables,
2002.
Statistics South Africa, 2008. LabourForce Survey 2007: Statistical Release P0210.
Su, B., Ang, B.W., 2014. Input–output analysis of CO2 emissions embodied in trade:
a multi-region model for China. Appl. Energy 114 (0), 377–384.
Su, B., Ang, B.W., Low, M., 2013. Input–output analysis of CO2 emissions embodied in
trade and the driving forces: processing and normal exports. Ecol. Econ. 88, 119–125.
United Nations, 1999. Handbook of Input–Output Table Compilation and Analysis,
Series F, No. 7. Department for Economic and Social Affairs Statistics Division.
United Nations, 1993. System of National Accounts 1993.
Van Der Linden, J.A., 2001. The economic impact study of maritime policy issues:
application to the German case. Marit. Policy Manag. 28 (1), 33–54.