View
5
Download
0
Category
Preview:
Citation preview
1
ECONOMIC GEOGRAPHY OF
INDONESIA’S MANUFACTURING INDUSTRY:
Dynamics and Changing Spatial Patterns1
Mudrajad Kuncoro Professor in economics Faculty of Economics & Business Gadjah Mada University, Yogyakarta, Indonesia 55281 Phone: +62811254255 Email: profmudrajadk@gmail.com
ABSTRACT
A number of economists and business strategists have become more interested in spatial analysis,
however, very few has examined the industrial clusters in Indonesia using provincial and district
data simultaneously. Indonesia, as the world‟s largest archipelago and 4th world most populous
country, offers an excellent laboratory for testing both old and new industrial district theories.
This article attempts to illuminate the clustering of Large and Medium Establishments (LME)
industry in Indonesia and, hence, enrich our understanding of not only the regional development
but also the relationship between urbanization and industrial development. Using Geographic
Information System (GIS) and Moran‟s I, we found hot and cold spots of industrial clusters.
Unlike the majority of manufacturing in US and UK that has concentrated in a relatively small
part of the country, within so-called the US manufacturing belt and UK‟s Axial Belt of industry
respectively, industrial clusters in Indonesia have concentrated geographically in Java‟s
metropolitan regions and major cities in the Outer islands. Our discriminant model suggests that
the best predictor of clustering was population density, followed by income per capita, skilled
workers, revenue sharing funds, infrastructure, productivity, and wages. Our study confirms that
the LME industrial clusters are strongly associated with urban concentration.
Key words: industrial cluster, Indonesia, GIS, LME, discriminant, geography
JEL classifications: R12, J24, J61
1 This paper is presented in the “Seminar Nasional & Kongres ISEI XIX”, Hotel Bumi Surabaya, 7-9 October 2015.
Earlier version of this paper has been presented in the Fourth Global Conference on Economic Geography 2015,
Oxford University, Oxford UK, 19-23 August 2015.
2
I. Introduction
With 241 million inhabitants ranked the 4th most populous country in the world in 2012 (BPS,
2013), Indonesia is the world‟s largest archipelago and offers an excellent laboratory for testing
both traditional and new industrial district theories. Since 1969, Indonesia‟s economic
development has been accelerated and the national goal of „Unity in Diversity‟ (Bhinneka
Tunggal Ika) has been accomplished. During the Soeharto era, hyperinflation was reduced
substantially into one to two-digit inflation, a high rate of capital flight was reversed into
substantial private capital inflows, foreign exchange reserve deficits were brought into positive,
self-sufficiency level in rice production was achieved in mid-1980s, sustainable economic growth
was generated, and poverty was reduced substantially (for example, Booth, 1992; Hill, 1996;
Kuncoro, et al. 1997). In the period 1997-1999, the economic crisis resulted in an economic
downturn of 2.9% per year, with the peak -13.1% in 1998. In the period 2000-2004, known as the
period of economic recovery, the economy had a positive growth of 4.5%. Within the period
2005-2008, the economic growth has reached an average of 6%. The Indonesian government
quite effectively steered the national economy through „storm‟ of the 2008 Global Financial
Crisis (Kuncoro, et al. 2009). Among the ASEAN-5, the Indonesian economy has proved to be
remarkably resilient, with output growing at 4.5% in 2009 compared with 1.75% for the ASEAN-
5 as a whole, thanks to strong domestic demand (Kuncoro, 2013).
This remarkable record in economic performance has been supported by the emergence of
the manufacturing sector as a leading sector in the economy. Figure 1 shows the declining role of
agriculture to GDP has been replaced by the increasing share of industry, especially the
manufacturing sector. In 1965, the agricultural sector had the dominant share of GDP (56%),
while the share of industry, including manufacturing, was only 13% of GDP. Indeed, with an
average of annual growth of 11.9% and 6.1% during 1965-1980 and 1980-1994 respectively,
industry has replaced agriculture as the leading sector in the Indonesian economy. In 1994, the
industry contributed 41% to GDP, in which the manufacturing sector had played a significant
role since it contributed 24% to GDP. At the same time the contribution of agriculture dropped
drastically to only 17% of GDP, while its growth declined steadily from 4.3% per annum during
1965-1980 to 3.1% during 1980-1994. Furthermore, the manufacturing sector has played an
important role in non-oil exports since it contributed to about 30% of non-oil GDP growth and
about three-fourths of non-oil export growth during 1986-1992 (Abimanyu, 1996: 26).
Nevertheless, manufacturing industry has grown below the Indonesia‟s economic growth since
2005.
Figure 1. Percentage share of agriculture and manufacturing industry to GDP: Indonesia, 1960-2011
Source: Calculated from BPS, various years
3
Given the increased importance of the industrial sector to Indonesia‟s economy, it seems
important to understand more fully patterns of industrial development, particularly its spatial
patterns. Mainstream economics has made only limited progress toward integrating spatial
considerations into its analytical corpus (Fujita, et al. 1999; Krugman, 1995; Krugman, 1998;
Ohta and Thisse, 1993). There are only few rigorous studies that explore spatial analyses of
industrial development in Indonesia, albeit with important exceptions (for example, Aziz, 1994;
Dick, et al. 1993; Henderson and Kuncoro, 1996; Hill, 1990; Kuncoro, 2000b). In addition,
Indonesian diversity in natural resources, population density, size of areas, and infrastructures
makes it imperative to take into account spatial considerations in industrial and regional
development. The lack of spatial consideration, together with the uneven development and an
unstable central government, resulted in serious problems of regional insurrection during the
1950s (Hill, 1989; Kuncoro, 1996b, 2000c).
The major purpose of this article is to illuminate the regional clustering of Large and
Medium Establishments (LME) of manufacturing industry in Indonesia and, hence, enrich our
understanding of the regional development. LME in Indonesia consist of large establishments
employing 100 workers or more, while medium establishments hiring 20 to 99 workers.
Manufacturing industry is an economic activity involving processing materials and transforming
them mechanically, chemically, or manually into finished or semi finished products and/or
converting them into other goods having higher value and closer to the final user (BPS, 2015).
The distribution of manufacturing establishments over space tends to cluster in certain
Indonesia‟s regions. We will show that LME manufacturing sector in Java and the Outer Island
has tended to cluster in and around major cities and towns. Manufacturing firms in Java seek to
locate in more populous and densely populated areas to enjoy both localization economies, which
are associated with the size of a particular industry, and urbanization economies, which reflect
the size of market of a district in a particular urban area (for example, Henderson, 1988; Kuncoro,
2000a). Manufacturing industries tend to concentrate geographically because agglomeration
economies have the effect of reducing cost or increasing sales for many types of industry. An
agglomeration represents not only a densely industrialised and urbanised region with diverse
industrial structure, but also amplified by agglomeration economies stemming from provision of
infrastructure services and public goods (Scott, 1993: 25-27). Manufacturing firms have tended to
locate their sites in and around urban regions as the regions offer various advantages in terms of
higher productivity and incomes that attract new investment, new technology, educated and
skilled workers to a disproportionate degree (Malecki, 1991). For the proponents of the new
economic geography, intersectoral linkages confined within these regions, together with thick
labour market, represents one of the key centripetal forces that tend to pull population and
production into agglomerations (Fujita and Thisse, 1996; Krugman, 1996: 7-10).
This paper attempts to analyze the dynamics of Indonesia‟s LME manufacturing industry
using spatial analysis and industrial district theories. In line with a dramatic increase in research
on economic geography -- that is, on „where‟ economic activity occurs and „why‟ (Fujita et al.
1999), this paper will address the following fundamental questions: Where are the main locations
of LME manufacturing industries in Indonesia? Why they tended to concentrate geographically
in some metropolitan and urban regions?
II. Theoretical background: old vs new industrial clusters An industrial cluster was initially associated with the Marshallian industrial district. Alfred
Marshall (1919), first observed the disposition of certain kinds of industries to localise in specific
areas in England, Germany and other countries, defined an industrial district as a specialised
4
geographical cluster of production. Interestingly, he distinguishes between „manufacturing town‟
and an „industrial district‟ as follows (Bellandi, 1989; Marshall, 1919: 285):
„Almost every industrial district has been focused in one or more large towns. Each such
large town, or city, has been at first the leader in the technique of industry, as well as in
trade; and the greater parts of its inhabitants have been artisans. After a time factories,
requiring more space than was easily to be had where ground values were high, tended to
move to the outskirts of the city; and new factories grew up increasingly in the
surrounding rural districts and small towns.‟
Marshall highlighted the role of three types of external economy as giving rise to the industrial
district: the concentration of a mass of skilled workers, the proximity of specialist suppliers and
the availability of facilities for obtaining knowledge. The presence of a large pool of skilled
workers facilitates economies of labour supply. The proximity of specialist suppliers produces
economies of specialisation arising from an extended division of labour between firms in
complementary activities and processes. The availability of facilities for obtaining knowledge
enhances economies of information and communication via joint production, invention and
improvements in machinery, processes and general organisation.
The relevance of Marshall‟s work is reflected by the growing development of similar
ideas in the more recent work of the Italian industrial district. Becattini (1990), for example,
defines the industrial district as a „socio-territorial entity which is characterised by the presence
of both a community of people and a population of firms and both tend to merge‟. He tries to
explain the Italian experience, especially in a region called the Third Italy (that is Tuscany,
Emilia Romagna, and nearby regions), which have been perceived by many authors since the
early 1980s as a model for the competitive success of clusters of small firms. Some major
features of the structure of manufacturing in the Italian industrial districts are geographical
concentration, sectoral specialisation, and predominance of strong network of small firms. These
characteristics have been observed in Silicon Valley (US), West Jutland (Denmark), and Baden-
Wurttemberg (Germany), Madrid, Fuenlabrada, Castellon, Mondragon and Valles Oriental
(Spain) (Pyke and Sengenberger, 1992); and some cases from Africa, Asia and Latin America
(Nadvi and Schmitz, 1994). However, a closer reading of those case studies reveals that there are
substantial differences between them in terms of their origins and consolidation as industrial
districts. Moreover, the applicability of the Italian model remains uncertain because of its highly
specific socio-historical context, mainly culture and institutions (Zeitlin, 1992).
Traditionally industrial districts have been centres of craft industry. Craft based, design-
intensive industries such as clothing, textiles, furniture, jewellery, ceramics, sporting goods, and
so forth, are found in old centres of craft production such as the Third Italy, parts of France,
Greece, Portugal, Germany, Spain, and Scandinavia (Scott and Storper, 1992). They may also be
found in the inner city areas of large metropolitan regions such as New York, Paris, Los Angeles,
and London. Modern-high technology clusters also emerge, such as ribbons, hardware, and
specialty steel in St Etienne; edge tools, cutlery, and specialty steel in Solingen, Remscheid and
Sheffield; and electronics in Southern California. Although a cluster is often characterised by a
particular industry, this may incorporate various „sub-industries‟. The obvious example is the
electronics manufacturing cluster in Southern California with multi-faceted focus on computers,
military and space communications equipment and avionics, and a diversity of components from
printed circuit boards to advanced semiconductor devices (Scott, 1993: chap.7).
A notable case discussed in the literature is the Brazilian supercluster in Sinos Valley. In
the 1960s it was a specialised cluster of craftsmen. Through success in export markets, by the
5
1990s it had become an internationally competitive industrial complex with flexible
specialization between firms of different sizes, from small (less than 20) to very large (over 500
workers). Its emergence as a supercluster reflected increased depth and density of the local
economy, which now includes almost the entire range of supply industries and producer services.
Besides four hundred and eighty shoe manufacturers, there are 1,341 related firms in service
industries: workshops, tanning industry, leather and footwear machine industry, component
industry, rubber industry, leather article industry, export and forwarding agents. Two causes of
this impressive development have been identified (Schmitz, 1995: 14). First, local and central
government incentives yielded quick results because export agents had connected Brazilian
producers with the United States market. Second, export growth increased the demand for local
inputs and machinery, thus contributing to development of the cluster.
Since the 1980s a post-Fordist literature has drawn attention to the flexibility (flexible
specialisation) and dynamism (employment generation) of industrial districts. As in the case of
the Brazilian supercluster, the elements of a cluster can be linked in three ways: (1) vertically or
convergently, when different stages of a process are involved, as in the case of spinning or
weaving or where assembly lines are fed by different sub-processes; (2) laterally, where the same
stage in a like process is involved, as in the case of men‟s clothing and women‟s clothing; and (3)
diagonally, when service processes are involved, such as repairing, trading, collecting, and so
forth. (Bellandi, 1989; Florence, 1961). Empirical studies have identified as major reasons for
the growth of industrial clusters and districts: (1) concentrations of highly specialised knowledge,
inputs and institutions, (2) the incentives of local competition, (3) the presence of sophisticated
local demand for a product or service, (4) geographical, cultural, and institutional proximity, and
(5) a proliferation of many different producers, all locked together in mutual interdependence
through their transactional relations (Porter, 1994, 1998a; Scott, 1993: chap.2).
The new form of industrial district derives from the industrial complex model, which
emerged from classical and neoclassical economics. The main features of the industrial complex
model are: (1) sets of identifiable and stable relations among firms which are conceived primarily
in terms of trading links; (2) minimisation of spatial transaction costs (that is transport costs,
telecommunication costs, shipment costs) in the formation of crucial, pre-planned or identifiable
linkages (Gordon, 2000). This model is in line with the Markusen‟s and Whittaker‟s study
suggesting the importance of vertical inter-firm linkages between LME and small and cottage
establishment (SCE) of “new” industrial districts in the USA and Japanese urban regions
(Markusen, 1996; Whittaker, 1997). A study of industrial districts in Japan shows the large
concentration of very small firms in the metropolitan centers of Tokyo, Osaka, and Nagoya
(Whittaker, 1997). Unlike hundreds of small firm clusters called sanchi, which produce traditional
or semi-traditional goods and tend to locate outside the main urban centres, Japanese metropolitan
concentrations are remarkable both for their scale and the high proportion of small firms with
their localised industry (jiba sangyo).
Building upon the pioneering work of Porter (1990) on competitive advantage, recent
research has begun to link the literature on sub-national clusters with that on firm strategy and
innovation (Enright, 1998; Porter and Solvell, 1998). The new theory of the firm with its
emphasis upon transaction costs and the transaction environment has great power to elucidate the
location decision of the firm. Region-specific resources, information flows, innovation,
technological spillovers and the formation of industry-specific skills can thereby be studied
within a consistent framework that has regard to both efficiency and dynamic effects (Porter and
Solvell, 1998: 441-2, figure 19.1). Porter (2003) found that the performance of regional
6
economies is strongly influenced by the strength of local clusters and the vitality and plurality of
innovation.
Unlike majority of previous studies, this study used combination of regional spatial data
and discriminant model to identify changing spatial patterns of LME manufacturing industry with
special reference to Indonesia. This study will fulfill the gaps, especially studies on industrial
districts and regional development for Indonesia using a discriminant model. Unlike most
previous studies that have discussed Indonesia‟s regional development at the provincial level
(e.g. Aziz 1994; Hill 1989; Manning 1997), this study will examine changing spatial patterns of
Indonesia‟s industrial development at the district (kabupaten/kodya) as well as provincial level.
Every study, which relies on provincial data, has at least two major shortcomings (Kuncoro
2000b: 11, 2012). First, provincial data tend to obscure urban economies. The province is either
too large or too small a unit. Even such large cities as Bandung, Semarang and Surabaya cannot
be seen in provincial data. Whereas the capital city region of Jakarta actually spills over its
provincial borders. Second, provinces are a much too large unit of analysis for any spatial
analysis of agglomeration economies (Gelder 1994: 64) as the heterogeneity within each province
would be too great. The disaggregation into districts, and even at the lower level (subdistrict or
kecamatan), enables us to understand the dynamic of regional clustering in some districts across
provinces.
III. Methodology
Spatial concentration of economic activities within a country indicates that
industrialisation constitutes a geographically selective process. Within the USA, for illustration,
the majority of manufacturing has been concentrated in a relatively small part of the country,
within the so-called manufacturing belt, since the second half of the nineteenth century
(Krugman, 1991: 11-4). Spatial concentration is also found in the UK‟s Axial Belt of industry
and the manufacturing belt of German Ruhr (Hayter, 1997: 45). This is also found in the case of
Indonesia, and to much extent the island of Java. Despite the fact that regional clusters are
usually associated with large cities, clusters can be also located in small and medium-sized
towns. As illustration, in Massachusetts a plastics cluster lies in the western part of state, jewelry
on the Rhode Island border, fiber optics around Sturbridge, textile and apparel near New
Bredford and Fall River (Porter 1994, 1998b).
This section will elaborate upon the methodology to identify LME industrial clusters. It
also attempts to illuminate on which regions the LME manufacturing industries tend to cluster in
Indonesia. The location of industrial clusters at the district (kabupaten/kodya) level will be
examined by using the Geographic Information System‟s (GIS) approach and Industrial Survey
data collected annually by Central Bureau of Statistics of Indonesia (BPS). The GIS is a useful
tool to identify the location of industry and where the manufacturing industries tend to cluster
(Kuncoro, 2001). The Industrial Surveys provide not only regional distribution of manufacturing
establishments but also their dynamics from 1976 to 2010. The surveys provide the plant-level
data of LME manufacturing firms, with more than 20 workers, that can be disaggregated by
industry code (ISIC) and district, providing all data of industry-specific variables.
A growing awareness of the limited explanations offered by the traditional location theory
has led to the surge of new economic geography, namely geographical economics (Kuncoro,
2000a; 2012). A growing number of economists have become interested in the study of location
problems (for example, Krugman, 1995; Lucas, 1988; Fujita, et al. 1999), which triggered a new
tool which has made an interesting contribution to geographical economics.
7
One of the major trends in the economic geography is employing the GIS in a socio-
economic context that focuses on the processes acting spatial data in its passage through
information system. A GIS is a special type of information system, concerned with the
representation and manipulation of a geographic reality. GIS transforms data into information by
integrating different data sets, applying focused analysis and providing output, all in manner to
support decision making (Juppenlatz and Tian, 1996: chap.1). The inventory, analysis, mapping
and modelling capabilities of GIS inevitably has wide application in a range of discipline,
ranging from information technology to socio-economic or population-related data (Martin,
1996: 4-5). Application of GIS in Indonesia has become more widespread (East Asian Executive
Reports 1996; Juppenlatz and Tian 1996: 103-108). In this study, we follow some typical
procedures involved in creating and using GIS. It includes data acquisition, preliminary data
processing, database construction, spatial search and analysis, and graphical display and
interaction (Kuncoro, 2012).
To identify industrial clusters, we apply the following steps. First, we rank all districts in
Indonesia for LME industry by employment and value added. Ranking all districts in terms of
both employment and value added in LME manufacturing industry reveals that these
establishments are unevenly distributed across Indonesia‟s regions. Figure 2 and 3 show that the
distribution of employment and value-added by districts is skewed rather than normal
statistically. The positive skew of the histograms indicates that there are some districts possess
high industrial density (in terms of employment and value added), while most contain very low
industrial density.
Figure 2. Employment distribution by district in Indonesia 2001 and 2010
Source: Calculated from BPS
Second, we map the employment and value-added data of LMEs to show where the
industrial and non-industrial clusters are located. For mapping purposes, a kotamadya (city) is
included in adjacent kabupaten or municipality (for example, kotamadya Semarang in kabupaten
Semarang). In this step, we differentiate between industrial and non-industrial districts by setting
a certain criteria (that is very high, high, moderate, and low) based on the total employment and
value added of each district (kabupaten/kotamadya). Industrial clusters are identified by either
“high” or “very high” industrial density in terms of both employment and value added. By
employing these criteria simultaneously, most of the top districts with „high‟ or „very high‟ total
employment and value added are found predominantly in Java, particularly in and around its
major cities namely Jakarta, Surabaya, Bandung, Semarang, and Surakarta. The same pattern
also occurs in the Outer Java Island, as most LME establishments are found in and around main
8
population centres such as Medan, Batam, Samarinda, Palembang, Pontianak, Ujung Pandang,
and Denpasar. The other kabupaten/kotamadya can be regarded as non-industrial areas since their
role in industrial employment and value added is minor.
Figure 3. Value added distribution by district in Indonesia 2001 and 2010
Source: Calculated from BPS
Empirical cluster research has contributed to the understanding the process of cluster
formation. Some experiences with the use of local spatial methods like local Moran‟s Ii and
Getis-Ord Gi tests in pattern recognition have been already available. Moran's I is a measure of
spatial autocorrelation developed by Moran (1950). In contrast, other measures, such as location
quotients, examine only the value for a single areal unit without reference to values in
neighboring areas. Our study utilises Moran‟s I to examine spatial autocorrelation that is
characterised by a correlation in a signal among nearby locations in space. Assessment of
industry location and density patterns becomes the first phase in the identification of potential
cluster regions to be included in a cluster driven development policy. Our study uses Getis-Ord
Gi in the identification of potential cluster regions in the Indonesia‟s LME industry.
One of major issues in economic geography: why do LME concentrate geographically in
some regions? More specifically, we will use discriminant model to show to what extent
Indonesia‟s foremost LME clusters represent industrial districts with distinctive features. If we
may classify those districts as a “separate” group from the rest of Indonesia, can we build a
predictive model, with group membership based on observed characteristics of each region?
Discriminant analysis is widely applied to serve the twin objectives of discrimination and
classification. Group separation is achieved by means of a discriminant function, while
identification of future individuals is handled through a classification rule (Krzanowski and
Marriott, 1995: 1). Discriminant analysis is a statistical technique for classifying individuals or
objects into mutually exclusive groups on the basis of a set of independent variables (Kuncoro,
2006; 2013). Unlike regression analysis, the objective in discriminant analysis is to find a linear
combination of the independent variables that minimise the probability of misclassifying
individuals or objects into their respective groups (Dillon and Goldstein, 1984: 360-3). The
overall test of relationship between predictors and groups in the discriminant analysis is the same
as the test of the main effect in multivariate analysis of variance (MANOVA), where all
9
discriminant functions are combined and grouping variables are considered simultaneously
(Tabachnick and Fidell, 1996). In other words, MANOVA allows us to look at how groups differ,
while discriminant analysis allows us to predict what variables discriminate between two or more
groups.
The discriminant function analysis was performed to predict the membership of
Indonesia‟s foremost LME industrial districts. To keep the analysis as simple as possible, we
dichotomise all districts into foremost industrial clusters versus the remainder as non-industrial
clusters. Modelling using discriminant approach is suitable given the data available. Our
discriminant function is based on the following equation:
Dik = a+ diX1+ diX2 + diX3 + diX4+ diX5 + diX6 + diX7 (1)
This study attempts to explore spatial discriminant patterns by contrasting discriminative
characteristics of distinct spatial entity classes using some predictors: productivity of labor
(X1), population density (X2), skilled workers (X3), income per capita (X4), infrastructure (X5),
revenue sharing funds (X6), and average wages (X7). We transform the predictors into natural
logarithm in order to normalise the distributions, which are positively skewed.
Productivity of Labour. Porter (1998) pointed out that industrial clusters, characterised
by geographic concentrations of interconnected companies and institutions in a particular field,
appeared to be far more productive industrial organizations than those based on one or two huge,
diversified cities. We define the productivity of labour as the ratio between value of output
and total employees. We test whether higher productivity of LME will influence the decision of
LME to cluster.
Population density. We use the population density by district as to show the effect of
generically named external economies of urbanization (Costa-Compi and Viladecans-Marsal,
1999: 2090). We will test whether a greater population density within a district corresponds to
greater probability that LME choose to locate in that region.
Skilled workers. Skilled workers are viewed by the Neo-classical school as a central
factor in the location decision of manufacturing establishments. A specific agglomeration
economy is that more highly educated and skilled labour concentrates disproportionately in main
cities. The theory of human capital is frequently applied to analyse the effect of education on
labour earnings (Tachibanaki,1998: 3). We will test whether higher skilled workers correspond to
greater probability that LME choose to locate in industrial districts.
Income Per Capita. We include Gross Regional Domestic Product per capita as a proxy
of market size. Krugman (1991: 23-4) argued that the more populated locations will attract a
concentration of manufacturing production, assuming that the location offers a sufficiently larger
local market than others, and fixed costs are large enough relative to transport costs. We will test
the Krugman‟s hypothesis.
Infrastructure. There is a substantial empirical literature showing that income and
production are correlated with market access in the way suggested by new economic geography
models. Fujita et al. (1999: 233-236) argued that a hub or a port city provide the city‟s site with
an advantage over other sites. A fundamental question in regional and urban economics is
whether lower transport costs cause agglomerations to develop, or if, instead, they promote the
dispersion of economic activity. In an infuential paper that started this literature, Krugman (1991)
shows the opposite: reducing transport costs causes agglomeration. Other theories have more
nuanced predictions, and the absence of sharp predictions demands credible empirical work. We
will test different theories of transport costs and economic geography, so that we can better
10
predict which regions will benefit from infrastructure improvements, and which regions may
suffer.
Revenue Sharing Funds. These funds are central government transfer to regions
reflecting resource endowments are agglomeration forces as argued by neo-classical economists,
such as Heckscher-Ohin (H-O). The H-O analysis establishes that “comparative advantage is
determined by the absolute distribution of resources between countries and particularly by the
relative factor endowment ratios between countries” (Johns, 1985: 178-81). East Kalimantan,
Riau, Aceh, Bangka Belitung are provinces that enjoyed higher revenue sharing funds because
they are known as resource rich regions (Figure 4). East Kalimantan, Aceh, and Riau are the
major producer of oil and gas in Indonesia. East Kalimantan is also home of coal mining and
forestry products. Bangka Belitung, as a new province splitting from South Sumatra, offers a
great tin mines and beauty of its nature based tourism. We will examine whether high revenue
sharing funds attract LME industry to locate in those regions.
Wages. Wages are one of the important variable in labour economics. Kuncoro (2007;
2012) found that wages are important in the textile, wood and miscellaneous industries but turn
out to be a less important variable in more modern LME industries such as machinery, chemical,
and paper. The story of wage trends in Indonesia, however, does not fit either with the modified
Lewis model or a neo-classical theories interpretation of sharply rising wage rates associated
with rapid economic growth (Manning, 1998: 114-131). Two features of the Indonesian labour
market transformation are relevant to our study: (1) the increasing share of non-agricultural,
urban and wage workers in total employment; and (2) a substantial increase in the number of non-
wage workers in urban areas (Manning 1998: 111-2). We include a labour cost variable
(WAGES), which is measured as wages for production workers, both male and female. We will
test whether the increase in wages will encourage LME to cluster in industrial districts.
Figure 4. Revenue Sharing Funds By Province
Source: Ministry of Finance (2013)
11
IV. Findings and Discussions
1. Where are LME industrial clusters? Most of Indonesian modern manufacturing establishments have persisted to cluster
predominantly on Java and to a much lesser extent, Sumatra Island during 1976-2010. Even when
we classify 33 provinces of Indonesia into five main islands (that is Sumatra, Java, Kalimantan,
Sulawesi, Eastern Islands), Java and Sumatra provided more than 93% of Indonesia‟s
employment and value-added over the period (Table 1). Java contributed about 75-89% of
Indonesian industrial employment and value added from 1976 to 2010, although this dropped
slightly during 1995-2001. Java‟s share declined from 89% and 86% in 1976 to 84% and 76% in
2010 in terms of industrial employment and value added respectively. Sumatra‟s share grew from
7 to 12% in terms of industrial emplyment and from 11 to 18% in terms of industrial value added
in the same period. Other main islands in Indonesia played a minor role in the Indonesia
manufacturing employment and value added. Even when we sum up the share of Kalimantan,
Sumatra, and Eastern Island, their share in Indonesian employment and value added was about
4% and 7% respectively over the period of 1976 to 2010.
Comparison between provinces in Java and the other islands in Indonesia is even more
pronounced. The share of each Java‟s provinces in terms of total employment and value added is
relatively higher than other provinces or islands in Indonesia from 1976 to 2010 (Table 2). Major
industrial regions in Java are found in its western and eastern part -- West Java, Jakarta, and East
Java --which contributed more than 50% of total employment and value added. This infers
relatively heavy regional clustering of manufacturing establishments in Java‟s vis-à-vis the rest of
the economy.
Table 1. Role of Java in the Manufacturing Employment and Value Added, Indonesia 1976-2010 (% of total)
Province/Island/Region Employment Value Added
1976 1985 1995 2001 2010 1976 1985 1995 2001 2010
Jakarta 15.2 13.8 10.7 9.06 6.94 25.7 17.9 17.9 14.29 14.27
West Java 19.1 23.3 35.2 26.99 28.20 19.9 25.3 33.8 20.67 27.82
Central Java and DIY 24.1 17.1 13.6 14.72 17.50 15.9 10.7 6.9 8,79 6,75
East Java 30.6 24.5 22.8 19.78 20.49 24.7 22.4 23.5 19.73 16.36
JAVA 89.1 78.6 82.2 82.04 83.73 86.2 76.3 82.1 74.35 75.73
SUMATRA 6.7 12.2 10.8 11.11 11.61 10.7 13.5 12.3 17.89 18.10
KALIMANTAN 1.8 5.6 3.9 3.73 2.17 1.8 6.7 4.0 4.59 4.12
OTHER EASTERN REGION 2.4 3.7 3.2 3.12 2.49 1.3 3.6 1.6 3.14 2.05
INDONESIA 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
Source: Calculated from BPS; Putro (2013)
There have been changing patterns in Java‟s employment for the last two decades (Figure
4). First, the declining share of Jakarta in the country‟s employment, albeit persistently
substantial in absolute value, has been offset partly by the rising share of West Java, East Java,
and Central Java‟s employment. In 1976 East Java was the top contributor to Java‟s employment,
followed by Central Java-DIY, West Java and DKI Jakarta (Recall Table 1). East Java has
sustained this record up till the mid of-1980s. However, a decade later the role of East Java
declined substantially and was replaced by West Java.
12
Table 2. Major Industrial Regions: Indonesia, 2001-2010
Extended Industrial Area
(EIA)
2001 2010
Employ-
ment (000) %
Value Added
(billion Rp) %
Employ-
ment (000) %
Value Added
(billion Rp) %
(1) (2) (3) (4) (5)
Java
2, 423 55.24 134.32 49.82 2.740 60,87 530.21 59.50
Jabodetabek 1,241 28.29 87.57 32.48 1.201 26.68 307.53 34.51
Bandung 562 12.82 19.73 7.32 561 12 .46 102.92 11.55
Surabaya 370 8.43 18.15 6.73 527 11.70 77.75 8.72
Semarang 158 3.59 6.22 2.31 359 7.97 33.05 3.71
Surakarta 92 2.11 2.65 0.98 92 2.05 8.97 1.01
Outer Java 265 6.05 21.04 7.8 289 6.42 57.42 6.45
Sumatera 202 4.61 17.94 6.65 251 5.58 52.12 5 .84
Medan 96 2.19 5.32 1.97 85 1.89 12.07 1.35
Batam 89 2.03 10.73 3.98 150 3.33 31.48 3.53
Palembang 17 0.39 1.89 0.70 16 0.36 8,57 0.96
Kalimatan 33 0.76 1.65 0.61 12 0.27 2.52 0.29
Samarinda 29 0.67 1.54 0.57 9 0.20 1.30 0.15
Pontianak 4 0.09 0.11 0.04 3 0.07 1.22 0.14
Sulawesi 19 0.42 1.08 0.40 19 0.41 2.38 0.27
Makasar 19 0.42 1.08 0.40 19 0.41 2.38 0.27
Bali 11 0.26 0.37 0.14 7 0.16 0.40 0.05
Denpasar 11 0.26 0.37 0.14 7 0.16 0, .40 0.05
Major Industrial Regions 2,688 61.29 155,36 57,62 3 67.28 587.64 65.9
INDONESIA 4,386 100 269,63 100 4.5 100 891.09 100
Source: Calculated from BPS; Putro (2013)
The second essential pattern is that almost every Java province showed a declining share
in the Indonesia‟s manufacturing employment. In 1976 DKI Jakarta, Central Java-DIY, and East
Java accounted for 15.2%, 24.1% and 30.6% of the Indonesian employment respectively. These
figures dropped substantially to be 10.7%, 13.6% and 22.8% in 1995. In contrast, West Java has
played a larger role in the country‟s employment since its share increased from 19.1% in 1976 to
35.2% in 1995, but declined to 28.2% in 2010.
13
Figure 4. Revenue sharing funds by province
Source: Ministry of Finance (2013)
The emergence of West Java and the declining share of DKI Jakarta in terms of
employment should be interpreted carefully. This is due to the fact that many industries have
expanded their activities around Jabotabek, an extended region which include Jakarta urban area
(DKI Jakarta) and its adjacent municipalities in West Java, namely Bogor and Tangerang
(Botabek). Rapid and massive industrial relocation has occurred as a lot of industries have moved
out of Jakarta to the outskirts to take advantage of lower costs and wages. The share of Jabotabek
employment located in Botabek rose from 43 to 56% between 1986 and 1991, while 73% of all
plant births in Jabotabek in the larger corporate sector occured in Botabek for 1989-1991
(Henderson, et al. 1996: 83-5).
In terms of value added, the picture is far more striking (Figure 5). When we combine
West Java and DKI Jakarta, this region dominates the Indonesian manufacturing value added.
These two provinces contribute more than 50% of the country‟s value added. However, a closer
look at the district level indicates that manufacturing establishments grow unevenly within West
Java and cluster particularly in and around Jabotabek and Bandung. We will show using GIS that
analysis using provincial data rather than municipality data blurs the dynamic of industrial spill-
over across administrative border. Therefore, the next section will analyse the Jabotabek region
as a whole rather than separate Jakarta and its surrounding municipalities (Botabek).
14
Figure 5. Percentage share of lme employment and value added in Java‟s provinces
Compared to Other Islands in Indonesia, 2001-2010
Source: Calculated from BPS
Outside Java the location of LME manufacturing tend to cluster in and around the
region‟s major cities and towns. At the provincial level, the major locations of LME
manufacturing firms in the Outer Island are found in North Sumatra, Riau, South Sumatra, West
Kalimantan, East Kalimantan, and South Sulawesi (recall Table 2). Known as either resource rich
or relative densely populated regions (for example, Hill, 1989, 1990), the distribution of LME
manufacturing sector in these provinces clusters heavily in a few districts (Table 4). In North
Sumatra most of the manufacturing firms --particularly aluminium, food, and plywood-- tend to
concentrate in Medan, Deli Serdang, and Asahan. In Riau basic metal products and plywood
firms tend to cluster in Kepulauan Riau, Bengkalis and Batam Island. In South Sumatra most
firms, particularly fertiliser, food and cement, are concentrated in Palembang and Musi Banyu
Asin. In West Kalimantan, timber and rubber processing firms are located in and around
Pontianak. Samarinda, Kutai, Bulungan, and Bontang are the major location of manufacturing
firms, particularly plywood and fertiliser, in East Kalimantan. In South Sulawesi, manufacturing
activities has located overwhelmingly in Ujung Pandang, and to lesser extent Pangkajene
Kepulauan.
In the Outer Island, large regional clusters are found in: (a) Batam Island and (b) Medan.
In terms of employment, both are much smaller than the main industrial areas in Java but
contribute substantially to value added. By value-added, Batam is slightly larger than Bandung
metropolitan regions but by employment it ranks only seventh (Kuncoro, 2012; 2004). Since the
mid-1980s Batam has developed rapidly in terms of infrastructure, investments, export, tourism,
and the manufacture of electronics (Smith, 1998: tables 18.1,18.2; Yuan, 1995: tables 12.1,12.2).
However, Batam is not an autonomous industrial agglomeration but an outlying industrial estate
of Singapore, just forty minutes away by ferry. National borders and differential regulations
allow land and labour to be much cheaper on Batam than on Singapore. As a product of both
policy and market forces, Batam, like Johor Baru in Malaysia, has become closely enmeshed with
Singapore as part of the Singapore-Johor-Riau (SIJORI) Growth Triangle. Its situation is quite
unlike that of most of the rest of the province of Riau, whether the adjacent islands or the main
island of Sumatra (Hill, 1996: 217). Batam must therefore be studied as part of the urban
economy of Singapore.
Medan has a long history as a centre for plantation industries such as rubber, tobacco and
oil palm, and its province of North Sumatra is still the most prosperous region in Sumatra
15
(Barlow and Wie, 1989). By employment and value-added, Medan regional clusters, including
Deli Serdang, achieves ranking with the main industrial areas of Java.
Other industrial areas in the Outer Island contribute only marginally in terms of
employment and value added. Palembang, whose role is significantly underestimated by
exclusion of oil refining, includes both heavy industry (fertiliser, cement) and resource
processing (tin, timber, and agro-industries). Samarinda‟s LMEs are narrowly specialised in the
timber industry, including saw mills, furniture, and plywood. Pontianak has a broader industrial
base consisting of crumb rubber, rubber-milling, coconut oil and paper industries, which are
nevertheless all resource-based. Other industrial areas in the Outer Island, such as Ujung
Pandang, are minor in terms of employment and value added.
Some major patterns with respect to industrial clusters in Indonesia can be drawn. First,
Indonesia‟s rapid industrial development under the New Order was heavily biased towards the
island of Java. The predominance activities in Java might be explained as follows. First,
historically the core of both the modern state and economy was moulded on Java during the
course of the nineteenth century, with the development of the plantation economy under the
Cultivation System (Dick, 1996). Second, the spatial pattern can be fairly readily explained by
the attraction of labour- and skill-intensive industries to heavily populated Java, compared with
the attraction of resource-based industries to the Outer Islands. In the Outer Island, regional
clusters can be associated with resource-based industry and urban attraction. The former is found
in the case of Asahan, Kepulauan Riau, Musi Banyu Asin, Bulongan, Kutai. The latter is
represented by the emergence of Medan and Batam, and to a lesser extent Samarinda, Ujung
Pandang, Palembang, Pontianak, and other cities.
Getis-Ord‟s Gi is proved useful in identifying clusters, particularly since it examines
patterns of LME co-location, or clusters, across areal unit boundaries within a specified
neighborhood district. Smith et al. (2003), for an example, has tested Getis-Ord Gi in the U.S
floriculture industry to measure spatial autocorrelation at the local level and identify “hot spots”,
or concentrations in spatial distributions in which areal units and their neighbors have similar
values of a given phenomena; a high Gi value indicates that high values are clustered near each
other, whereas a low Gi value is low spatial cluster (“cold spot”) indicating of low values being
near each other.
Figure 6 indicates that employment growth of LME (growthcap) has a positive value of
0.6525. It illuminates how spatial clusters of LME industry tended to concentrate on certain
districts. We may identify the hot and cold spot of LME clusters in Indonesia. Strong
concentrations in spatial distributions in which areal units and their neighbors, or hot spots, are
found in Banten, DKI Jakarta, Jawa Barat, dan Central Java (shown by red colour in Figure 7).
These regions have not only high growth of LME employment but also bring high regional
spillover. On the other hand, cold spot region in Indonesia is found in South Sulawesi (shown by
blue colour in Figure 7). It implies that the province has relatively low spatial clusters and
regional spillover.
16
Figure 6. Moran‟s I and Scatter Plot of GrowthCap Indonesia
Figure 7. Gi growth cap of LME clusters in Indonesia
Source: Calculated from BPS
2. Why LME tended to cluster in some regions?
To address this question, we test a discriminant model as specified in equation 1. Overall,
our discriminant model allocates correctly more than 80.7% of the original group cases. Table 3
Total Employment (TotalTK) Moran‟s I= 0,652548
(Spatial Clustering)
17
provides a classification summary for the model, which incorrectly allocates only 65 districts out
of 326 to non-industrial districts. In terms of industrial districts, the model fails to allocate just 10
cases. As a result, the correctly predicted group membership is 80.1% for non-industrial
districts and 84.1% for industrial districts.
Table 3 Classification resultsa
Dik
b Predicted Group Membership Total
0 1
Original Count 0 261 65 326
1 10 53 63
% 0 80.1 19.9 100
1 15.9 84.1 100
Note: a 80.7% of original grouped cases correctly classified.
b Dik is 1 for industrial district; 0 for non-industrial districts.
The results of a direct discriminant function analysis using seven predictors suggest
that population density play a key role as the best predictor in discriminating LME between
industrial districts and non-industrial districts (Table 4). The coefficient for this variable
shows a positive sign. This implies that the higher the population density in a district the more
likely that LME will cluster towards industrial districts. Figure 8 shows there are positive
association between industrial density and population density across regions in Indonesia.
Table 4 Discriminant function coefficients
Predictor Function
Population density 0.830
Income per capita 0.374
Skilled workers 0.222
Revenue sharing funds -0.170
Infrastructure 0.050
Productivity 0.044
Average wages -0.029
Pooled within-groups correlations between discriminating variables and standardised
canonical discriminant functions. Variables ordered by absolute size of correlation within
function.
Population density, regional income per capita, and skilled workers show positive
coefficients. The positive coefficient of population density and income per capita show that both
scale economies and large market size explain regional localisation over time that attract LME to
locate in the industrial clusters. The findings confirm the prediction of New Trade Theory and
New Economic Geography: scale economies and home market do matter (Kuncoro, 2012;
Krugman, 1991; Fujita and Venables, 1999). These results suggest that LME manufacturing
firms in Indonesia seek to locate in more populous and densely populated areas to enjoy both
localization economies, which are associated with the size of a particular industry, and
agglomeration economies, which reflect the size of market of a district, in a particular urban area.
18
Note:
High Urban Center concentration and High Population density
High Urban Center concentration and Low Population density
Low Urban Center concentration and High Population density
Low Urban Center concentration and Low Population density
Figure 8. Employment distribution by main islands and urban centers in Indonesia, 2010
Source: Calculated from BPS; Putro (2013)
Skilled workers, reflecting the availability of skilled workers, played an important role of
LME clustering. The positive coefficient of skilled workers shows that the higher level of labour
force education the higher possibility of LME to locate in the industrial clusters. The finding is
line with Smith and Florida (1994) which examined that big cities has become a magnet for
skilled workers to concentrate geographically in urban regions to seek jobs.
Revenue sharing funds as a central government transfer to regions reflect agglomeration
forces through resource based industry as argued by neo-classical economists (Johns, 1985: 178-
81). The negative coefficient of revenue sharing funds suggest that higher revenue sharing
funds the lower the possibility to attract LME industry to locate in those regions. Known
as rich resource based regions, East Kalimantan, Riau, Aceh, Bangka Belitung are provinces that
enjoyed higher revenue sharing funds. However, only mining companies, rather than LME
manufacturing, that clustered in those provinces.
The infrastructure and productivity are also found as good predictors. The positive sign of
infrastructure coefficient suggests that the better infrastructure availability in a region the greater
possibility of LME to cluster in the region. The theoretical and empirical literature on
agglomeration economies and urban public infrastructure pinpoints that agglomeration economies
exist when firms in the urban areas share public goods as inputs to production (Eberts and
McMillen, 1999). The finding is in line with Fujita et al. (1999: 233-236) and Krugman‟s (1991)
argument. The former argued that a hub or a port city provide the city‟s site with an advantage
over other sites, while the latter shows that reducing transport costs causes agglomeration.
Keterangan:
19
Likewise, the positive sign of productivity coefficient means that the higher LME
productivity will lead the LME to cluster in and around industrial districts. This findings support
Porter‟s (1998) theory that industrial districts constitute more productive industrial organization
than non-industrial districts.
Wages are one of the important variables in labour economics but play as the least predictor
in discriminating between LME industrial districts and non-industrial districts. The negative
coefficient of wages shows that the higher wages the lower the possibility to attract LME
industry to locate in those regions. We found that increase in wages will discourage LME to
cluster in industrial districts. This finding supports Kuncoro‟s (2007, 2012) argument that wages
are important in the textile, wood and miscellaneous industries but turn out to be a less important
variable in more modern LME industries such as machinery, chemical, and paper. Our findings
suggest that higher wages in the industrial districts will encourage LME to expand
and seek other districts offering lower wages.
V. Conclusions
Although a number of economists and business strategists have recently become more
interested in spatial analysis, very few has examined the phenomena of regional clustering in
Indonesia using provincial and district data simultaneously. Some studies focus on urban
agglomeration, while others relate clustering to the competitive advantage of firms (Anselin and
Florax, 1995; Enright, 1998; Hill, 1997; Krugman, 1995; Lucas, 1988; Porter and Solvell, 1998).
This growing awareness on the spatial aspects of economies, the theoretical insights that have
been developed are as yet little tested empirically.
Our study sought to inject empirical content to the emerging interest in the literature by
examining regional clustering in the context of Indonesia‟s recent industrial transformation.
Using a GIS and recent perspectives in the literature of cluster, our study demonstrates that
Indonesia represents an excellent example of both the uneven geographic distribution of
manufacturing industry and the relationship between urbanization and industrial development.
Indonesia contains many regional clusters within a fairly homogeneous economic and regulatory
environment.
Indonesia‟s rapid industrial and regional development during the last two decades has
been heavily biased towards the densely populated island of Java. This is most striking in the
case of LME manufacturing, corresponding with the modern sector of manufacturing. The
predominant activities in Java can be explained as follows. First, historically during the course of
the nineteenth century, the core of both the modern state and economy was moulded on Java,
with the development of the plantation economy under the Cultivation System (Dick, 1996).
Second, the spatial pattern can be quite readily explained by the attraction of labour- and skill-
intensive industries to the heavily populated Java, compared with the attraction of resource-based
industries to the Outer Islands (Kuncoro, 2012; 2013).
This study confirms that the LME industrial clusters are associated with urban
concentration in Indonesia. Important distinguishing features of the industrial district have been
identified by a discriminant analysis. The results suggest that the best predictor was population
density, followed by income per capita, skilled workers, revenue sharing funds, infrastructure,
productivity, and averages wages. Thus, LME in industrial districts differ statistically from the
rest of Indonesia in terms of those key variables.
The Indonesian government has given attention to clustering perspective in the new
National Industrial Policy and MP3EI recently (Kuncoro, 2007, 2013). As pointed out by
Minister of Industry and Trade, the benefit of industrial clusters is to encourage product
20
specialisation and to change comparative advantage into competitive advantage (Kompas, 2000;
Kuncoro, 2012). In line with this policy direction, understanding the locations, causes, and
consequences regional clustering is imperative. This study serves as a valuable input to the
government in formulating the new direction for industrial policy.
References
Abimanyu, A. (1996) The Indonesian economy and total factor productivity. The Singapore
Economic Review, 40(1), 25-40.
Anselin, L., and Florax, R. J. G. M., eds. (1995) New Directions in Spatial Econometrics.
Springer: New York.
Aziz, I. J. (1994) Ilmu Ekonomi Regional Dan Beberapa Aplikasinya Di Indonesia [Regional
Economics and its Some Applications in Indonesia], Lembaga Penerbit Fakultas Ekonomi
Universitas Indonesia: Jakarta.
Barlow, C., and Wie, T. K. (1989) North Sumatra: Growth with unbalanced development, in H.
Hill (ed.), Unity and Diversity: Regional Economic Development in Indonesia since 1970
(pp. 409-36), Oxford University Press: Oxford.
Bellandi, M. (1989) The Industrial District in Marshall, in E. Goodman, J. Bamford, and P.
Saynor (eds.), Small firms and industrial districts in Italy, Routledge: London and New
York.
Booth, A. ed. (1992) The Oil Boom and After: Indonesian Economic Policy and Performance in
the Soeharto Era. Oxford University Press: Singapore.
Dick, H., Fox, J. J., and Mackie, J. (eds). (1993) Balanced Development: East Java in the New
Order. Oxford University Press: Singapore.
Dick, H. W. (1996) The emergence of a National Economy, 1808-1990s. in J. T. Lindblad (ed.),
Historical Foundations of a National Economy in Indonesia, 1890s-1990s, 21-51, Royal
Netherlands Academy of Arts and Science: Amsterdam.
Djidin, D. A. (1997) The political economy of Indonesia's new economic policy. Journal of
Contemporary Asia, 27(1), 14-36.
Eberts, R. W., and McMillen, D. P. (1999) Agglomeration economies and urban public
infrastructure, in P. Cheshire and E. S. Mills (eds.), Handbook of regional and Urban
Economics: Applied Urban Economics, 3: 1457-1495. North-Holland: Amsterdam.
Enright, M. J. (1998) Regional clusters and firm strategy. in A. D. J. Chandler, P. Hagstrom, and
O. Solvell (eds.), The Dynamic Firm: The Role of Technology, Strategy, Organization,
and Regions, 3150-42. Oxford University Press: Oxford.
Fane, G. (1996) Deregulation in Indonesia: Two steps forward, one step back, Agenda, 3(3), 341-
50.
Florence, P. S. (1961) The Logic of British and American Industry. London: Routledge and
Kegan Paul.
Fujita, M. Krugman, P. and Venables, A. J. (1999) The Spatial Economy: Cities, regions, and
International Ttrade. The MIT Press: Cambridge and London.
Fujita, M., and Thisse, J.-F. (1996) The economics of agglomeration, Journal of Japanese and
International Economics, 10: 339-78.
Gee, T. G., and Robinson, I. M. (eds.) (1995) The Mega Urban Regions of Southeast Asia. UBC
Press: Vancouver.
21
Gelder, L. v. (1994) Industrial agglomeration and factor market segmentation with empirical
applications to Indonesia. Unpublished PhD dissertation, Cornell University, US:
Cornell.
Gilmour, J. M. (1974) External economies of scale, inter-industrial linkages and decision making
in manufacturing. in F. E. i. Hamilton (ed.), Spatial Perspectives on Industrial
Organization and Decision Making, 335-62. John Wiley and Sons: London.
Hardjono, J., and Hill, H. (1989) West Java: Population pressure and regional diversity, in H. Hill
(ed.), Unity and Diversity: Regional Development in Indonesia since 1970, 255-282.
Oxford University Press: Oxford.
Henderson, J. V. (1988) Urban Development,Theory, Fact, and Illusion. Oxford University
Press: New York.
Henderson, J. V., and Kuncoro, A. (1996) Industrial centralization in Indonesia, The World Bank
Economic Review, 10(3), 513-40.
Henderson, J. V., Kuncoro, A. and Nasution, D. (1996) The dynamic of Jabotabek development,
Bulletin of Indonesian Economic Studies, 32(1), 71-95.
Hill, H. (ed.) (1989) Unity and Diversity: Regional Economic Development in Indonesia Since
1970. Oxford University Press: Singapore.
Hill, H. (1990) Indonesia's industrial transformation Part 1, Bulletin of Indonesian Economic
Studies, 26(2), 79-120.
Hill, H. (1996) The Indonesian Economy Since 1966: Southeast Asia's Emerging Giants.
Cambridge University Press: Cambridge.
Hill, H. (1997) Indonesia's Industrial Transformation. Institute of Southeast Asian Studies:
Singapore.
Jong, T. D., and Floor, H. (1996) Analysis of interaction data in a GIS environment-a case study
in Indonesia, in T. Fung, P. C. Lai, H. Lin, and A. G.-O. Yeh (eds.), GIS in Asia,chap. 32.
GIS Asia Pacific: Singapore.
Juppenlatz, M., and Tian, X. (1996) Geographic Information Systems and Remote Sensing.
McGraw-Hill Book Company: Sydney.
Kompas. (2000) Kebijakan nasional sektor industri: Aglomerasi dengan kemitraan {The national
policy of industrial sector: Agglomeration with partnership], Kompas. Available:
http://www.kompas.com.kompas-cetak/008/19/EKONOMI/aglo14.html [2000, 19
August].
Krugman, P. (1991) Increasing returns and economic geography, Journal of Political Economy,
99, 483-499.
Krugman, P. (1995) Development, Geography, and Economic Theory. The MIT Press:
Cambridge and London.
Krugman, P. (1996) Urban concentration: The role of increasing returns and transport costs,
International Regional Science Review 19(1 and 2), 5-30.
Krugman, P. (1998) Space: The final frontier, Journal of Economic Perspectives 12(2), 161-74.
Kuncoro, M. (2000a) Beyond agglomeration and urbanization, Gadjah Mada International
Journal of Business, 2(3).
Kuncoro, M. (2000b) The Economics of Industrial Agglomeration and Clustering, 1976-1996:
the case of Indonesia (Java). Unpublished PhD thesis, the University of Melbourne:
Melbourne.
Kuncoro, M. (2000c) Ekonomi Pembangunan: Teori, Masalah dan Kebijakan. UPP AMP
YKPN: Yogyakarta.
22
Kuncoro, M. (2001) Regional clustering of Indonesia‟s manufacturing industry: a spatial analysis
with Geographic Information System (GIS), Gadjah Mada International Journal of
Business, 3 (3).
Kuncoro, M. (2006) Quantitative Methods: Theory and Applications for Business and
Economics, 3rd
ed., UPP STIM-YKPN: Yogyakarta.
Kuncoro, M. (2011a) Koridor ekonomi Indonesia [Economic corridors of Indonesia], Investor
Daily, Retrieved February 28, 2011, from http://website.mudrajad.com/content/koridor-
ekonomi-indonesia.
Kuncoro, M. (2011b) MP3EI: mitos atau realitas? [MP3EI: a myth or reality?], Retrieved August
1, 2011, from
http://cetak.kompas.com/read/2011/08/01/0415106/mp3ei.mitos.atau.realitas.
Kuncoro, M. (2012) Ekonomika Aglomerasi: Dinamika dan Dimensi Spasial Kluster Industri
Indonesia [Economics of Agglomeration: Dynamics and Spatial Dimension of
Indonesia’s Industrial Clusters]. UPP STIM YKPN: Yogyakarta.
Kuncoro, M. (2013) Economic Geography of Indonesia, South East Asian Journal of
Comtemporary Business, Economics, and Law, 2(2), 17-33.
Kuncoro, M., Adji, A. and Pradiptyo, R. (1997) Ekonomi Industri: Teori, kebijakan, dan Studi
Empiris di Indonesia [Industrial Economics: Theory, Policy, and Empirical Studies in
Indonesia]. Widya Sarana Informatika: Yogyakarta.
Kuncoro, M, Widodo, T. and McLeod, R. (2009). Survey of recent developments, Bulletin of
Indonesian Economic Studies, 45(2), 151-176.
Lucas, R. E. (1988) On the mechanics of economic development, Journal of Monetary
Economics, 22, 3-22.
Malecki, E. J. (1991) Technology and Economic Development: The Dynamics of Local, Regional,
and National Change. John Wiley and Sons, Inc: New York.
Manning, C. (1997) Regional labor markets during deregulation in Indonesia: Have the outer
islands been left behind?, Policy Research Working Paper, No.1728. The World Bank,
East Asia and Pasific, Country Department III: Washington, DC.
Markusen, A. (1996) Sticky places in slippery space: A typology of industrial districts, Economic
Geography, 72(3), 293-.
Marshall, A. (1919) Industry and Trade. Macmillan: London.
Martin, D. (1996) Geographic Information System: Socioeconomic Applications, 2nd ed.
Routledge: New York.
Ministry of Finance. (2013) Data Pokok APBN 2007–2013, Retrieved January 31, 2014, from
http://www.anggaran.depkeu.go.id/dja/acontent/Data%20Pokok%20APBN%202013.pdf.
Moran, P. A. P. (1950) Notes on continuous stochastic phenomena, Biometrika, 37(1): 17–23.
Nadvi, K., and Schmitz, H. (1994) Industrial clusters in LDCs: Review of experiences and
research agenda, IDS Discussion Paper No.339. IDS, University of Sussex: Brighton.
Nakamura, R. and Sendouw, R.H.E.(2009) Market potential and regional per capita GRP
differentials: application of the NEG model to Indonesian provinces. Retrieved June 14,
2013, from
http://www.researchgate.net/publication/32199787_Market_Potential_and_Regional_Per_
Capita_GRP_Differentials__Application_of_the_NEG_Model_to_Indonesian_Provinces.
Nasution, A. (1991) Recent economic reform, The Indonesian Quarterly, 29(1), 12-26.
Ohmae, K. (1995) The End of the Nation State: The Rise of Regional Economies. London: Harper
Collins.
23
Ohta, H., and Thisse, J.-F. (eds.) (1993) Does Economic Space Matter? Essays in Honour of
Melvin L. Greenhut. St. Martin's Press: New York.
Porter, M. E. (1990) The Competitive Advantage of Nations. The Free Press: New York.
Porter, M. E. (1994) Comment on interaction between regional and industrial policies: Evidence
from four countries, by Markusen. Proceedings of the World Bank Annual Conference on
Development Economics 1994, Washington DC.
Porter, M. E. (1998a) Clusters and the new economics of competition, Harvard Business Review,
November-December (6), 77-91.
Porter, M. E. (1998b) Competing across locations: Enhancing competitive advantage through a
global strategy, in M. E. Porter (ed.), On Competition,309-48. Harvard Business School
Publishing: Boston.
Porter, M. E., and Solvell, O. (1998) The role of geography in the process of innovation and the
sustainable competitive advantage of firms, in J. Alfred D. Chandler, P. Hagstrom, and O.
Solvell (eds.), The Dynamic Firm: The Role of Technology, Strategy, Organization, and
Regions, 440-457. Oxford University Press: Oxford.
Putro, S.S. (2013) Analisis lokasi aglomerasi dan kluster industri besar sedang Indonesia Tahun
2001-2010 [A locational analysis of Indonesia's large and medium industrial clusters
2001-2010]. Master of Economic Development Program UGM: Yogyakarta. unpublished
Master thesis.
Pyke, F., and Sengenberger, W. (eds.) (1992) Industrial Districts and Local Economic
Regeneration. ILO: Geneva.
Schmitz, H. (1995) Small shoemakers and fordist giants: A tale of a supercluster, World
Development, 23(1), 9-28.
Scott, A. J. (1993) Technopolis: High Technology Industry and Regional Development in
Southern California. University of California Press: Berkeley.
Scott, A. J., and Storper, M. (1992) Regional development reconsidered, in H. Ernste and V.
Meier (eds.), Regional Development and Contemporary Industrial Response: Extending
Flexible Specialisation, 3-24. Belhaven Press: London.
Smith, B.W., Carroll, M.C., Reid, N. (2003) Potential cluster regions: The case of the floriculture
industry, http://uac.utoledo.edu/nwoerc/FloriculturePCRs.pdf. April 2, 2013.
Smith, S. L. (1998) Batam Island and Indonesia's high technology strategy, in H. Hill and T. K.
Wie (eds.), Indonesia's Technological Change, 342-363. Institute of Southeast Asian
Studies: Singapore.
Statistics Indonesia (BPS). 2015. Concept and methodology of Large and Medium
Manufacturing, http://www.bps.go.id/Subjek/view/id/9#subjekViewTab1|accordion-
daftar-subjek2. July 22, 2015.
Yuan, L. T. (1995) The Johor-Singapore-Riau growth triangle: The effect of economic
integration, in T. G. McGee and I. M. Robinson (eds.), The Mega-Urban Regions of
Southeast Asia, UBC Press: Vancouver.
Zeitlin, J. (1992) Industrial districts and local economic generation: Overview and comment, in
F. Pyke and W. Sengenberger (eds.), Industrial Districts and Local Economic
Regeneration, ILO: Geneva.
Recommended