WORKING PAPER

16-02

The New Economic

Geography:
a survey of the
literature

Federal
Planning Bureau
Economic analyses and forecasts

Avenue des Arts 47-49 D. Simonis

B-1000 Brussels
Tel.: (02)507.73.11
Fax: (02)507.73.73 December 2002
E-mail: contact@plan.be
URL: http://www.plan.be
The New Economic
Geography:
a survey of the
literature

D. Simonis

December 2002
Federal Planning Bureau
The Belgian Federal Planning Bureau (FPB) is a public agency under the authority
of the Prime Minister and the Minister of Economic Affairs. The FPB has a legal sta-
tus that gives it an autonomy and intellectual independence within the Belgian
federal public sector.

FPB’s activities are primarily focused on macroeconomic forecasting, analyzing and

assessing policies in economic, social and environmental fields.

Internet
URL: http://www.plan.be

E-mail: contact@plan.be

Publications
Publications:
The Economic Forecasts
The “Short Term Update”

Planning Papers (recent publications)

giving the results of policy analyses and assessments
91 Verkenning van de financiële evolutie van de sociale zekerheid 2000-2050 -
De vergrijzing en de leefbaarheid van het wettelijk pensioensysteem
M. Englert, N. Fasquelle, M.-J. Festjens, M. Lambrecht, M. Saintrain,
C. Streel, S. Weemaes - January 2002
92 De administratieve lasten in België voor het jaar 2000 - Eindrapport
Greet De Vil, Chantal Kegels - January 2002

Working Papers (recent publications)

12-02 Constructing Productive ICT Capital Stock Series for Belgium
T. Pamukçu, W. Van Zandweghe - October 2002
13-02 Stedelijke woondynamiek van de Belgische bevolking en haar gezinnen
D. Devogelaer - November 2002
14-02 Geographical agglomeration: the case of Belgian manufacturing industry
L. Bertinelli, J. Decrop - November 2002
15-02 Opleiding en kwalificaties, arbeidsmarkt en ICT: de rol van onderwijs en
training
Gijs J.M. Dekkers - November 2002

Responsible Editor Henri Bogaert

Legal Depot: D/2002/7433/33
 Working Paper 16-02

Table of Contents

Executive Summary 1

I Introduction 3

II The New Economic Geography:

background and main features 5

A. The resurgence of geography 5

B. Increasing returns to scale 6

C. Agglomeration economies 7

D. Cumulative or circular causation mechanism 8

E. Centripetal and centrifugal forces 8

III The Core-Periphery model 11

A. The spatial impossibility theorem 11

B. The modelling tools 11

1. Dixit-Stiglitz monopolistic competition framework 11
2. Transport costs 12

C. The standard Core-Periphery model 12

1. Assumptions 13
2. Conditions for agglomeration 13

D. The extentions of the Core-Periphery model 15

IV Cities and economic growth 17

A. Economic growth and agglomeration 17

B. Growth in cities 18
1. Knowledge spillovers 18
2. Agglomeration forces and congestion forces in cities 19
3. Optimal degree of concentration 20

C. ICT in growth and agglomeration 21

1. Some features of ICT 21
2. “The death of distance” 21
Working Paper 16-02

V Conclusion 23

Bibliographie 25
 Working Paper 16-02

Executive Summary

This overview of the literature dedicated to the new economic geography

intends to highlight the main mechanisms, which contribute to explain the spa-
tial concentration of economic activity, in particular the formation of cities and
industrial districts. This should provide some guidelines for an empirical analy-
sis of the determinants of the spatial distribution of economic activity in urban
areas in Belgium and for suggestions of economic policy instruments capable of
influencing location choices.

The main contribution of the new economic geography is to deal with some of
the classic questions of regional and urban sciences related to location in a coher-
ent theoretical framework, and to explain the endogenous mechanisms driving
geographic concentration of economic activity and leading to core-periphery
patterns. In a pioneering text, Krugman (1991a) has shown that these forms of
concentration are linked to the existence of agglomeration economies, according
to which the spatial concentration itself creates a positive economic environ-
ment.

According to the new economic geography, the self-reinforcing process deriving

from the spatial concentration of economic activity arises from the interplay of
various factors: scale economies, transport costs, backward-forward linkages of
firms, which determines the location of economic activity. In the presence of
increasing returns to scale, each producer is inclined to concentrate its produc-
tion in a single location. In order to minimise the transport costs, the producer
will have a tendency to locate where the local demand is the most important. But
demand concentrates precisely where other firms have already located. If a spe-
cific economic activity develops particularly in a given region, due to history,
this region will attract the firms of the other regions, thus reinforcing the advan-
tage deriving from the size of its own market. This circular process of cumulative
causality leads in the end to the concentration of the industry in a single area.

The effects from the concentration of economic activity can be reinforced by the
existence of externalities such as technology and knowledge spillovers through
the improvement of information flows (informal contacts facilitated by proxim-
ity), access to a diversified range of intermediate goods and complementary
services to industrial activity, and the benefits from specialised high-skilled
labour availability. The economic literature also makes a distinction between
location economies associated with the firms belonging to a same sector and the
urbanisation economies associated with the firms from all sectors being located
in a same place. The accounting for the interactions between technological, sec-
toral and geographical proximity comes from the research work on endogenous
growth, which considers these externalities as the engine of growth.

1
Working Paper 16-02

However, congestion effects exist besides these agglomeration economies.

According to these effects, concentration brings about an increase in the prices of
the immobile local factors (land, natural resources, agricultural workers), and of
goods that can lead to an inverse process of dispersion of economic activity.
These dispersion forces, which allow to explain the existence of several concen-
tration places, are linked to transport costs, land use costs, labour market compe-
tition and competition on the market of goods and services, and to negative
externalities (congestion in transport, pollution).

From an economic policy point of view, it is very important to understand the

processes underlying the location decisions of firms and their possible impact on
industrial structure and the dispersion of economic activity within a given geo-
graphical area, because the predominance of agglomeration forces could create
inequalities between the core and the periphery. The central idea of the new eco-
nomic geography is that the spatial distribution of economic activity results from
the interaction between centripetal forces and centrifugal forces, whose nature
can vary according to the spatial scale considered: local, regional or interna-
tional. While the role of pecuniary externalities, emphasised by the new eco-
nomic geography, is especially relevant to explain agglomeration effects at the
European scale, empirical studies tend to show that the existence of technology
and knowledge spillovers, encouraged by specialised high-skilled labour availa-
bility, may well prove to be a better explanation of agglomeration at the local
level. Thus, this mechanism is very important to understand, for example, the
impact of information and communication technology on the spatial configura-
tion of cities.

2
 Working Paper 16-02

I Introduction

The “New Economic Geography” deals with some of the classic questions of
regional and urban sciences. Why does economic activity tend to concentrate
geographically? How has the spatial distribution of economic activity evolved
and how can it be expected to evolve in the future? What is the appropriate role
of government in influencing this evolution? However, it tends to revisit these
questions using a new approach where increasing returns to scale are the driving
force behind increased concentration of economic activities.

An obvious explanation for the uneven distribution of economic activity across

space is that certain regions enjoy natural advantages over others, such as
endowments with natural resources or the presence of rivers or harbours. How-
ever, additional arguments have to be introduced in order to understand the
advantages of concentration, which are unrelated to natural endowments. The
new economic geography relies on the concept of “agglomeration economies”,
i.e. economies arising from the interplay of transport costs1, increasing returns to
scale and factor mobility, in order to explain the clustering of firms, workers and
consumers.

By introducing the concept of “circular causation”, i.e. the idea that dominance
of regions is a self-reinforcing process, the new economic geography models try
to explain how core-periphery patterns can arise endogenously. They help
understand how historical accident2 can shape economic geography, and how
small changes can produce discontinuous changes in spatial structure.

From an economic policy point of view, it is very important to understand the

processes underlying the location decisions of firms and their possible impact on
industrial structures and the dispersion of industrial activities within a geo-
graphical area. The central idea is that the population and production patterns
result from the interaction between centripetal forces (forward-backward link-
ages) and centrifugal forces (immobility of factors of production). The new eco-
nomic geography can be useful to define the conditions under which a
movement of firms toward the core can be expected and when this trend can
reverse. The predominance of agglomeration forces could extend the imbalance
between a core and a periphery and create spatial inequalities.

The geographic concentration of economic activity is a characteristic of the eco-

nomic landscape. Agglomerations can be observed at different scales.

1. In the new economic geography, as in international trade models, the term “transport costs” cov-
ers all sorts of costs that arise in international transactions apart from genuine transport costs:
tariffs, quotas, and information costs due to national regulations, culture, language, etc.
2. More generally, the idea that there is a “path dependency” and that “history matters” has been
previously developed by David (1985) and Arthur (1989, 1994).

3
Working Paper 16-02

• At the broader scale, the new economic geography analyses the factors,
which lead to the clustering of economic activities in regions such as the US
"manufacturing belt" (area between Green Bay, Saint Louis, Baltimore and
Portland) or the European manufacturing core, the so-called “Blue Banana”
(area between South East England, Ruhr Valley, South East France, Southern
Germany and Northern Italy). This deepening integration can lead to the
strengthening of the core regions, attracting new production sectors, at the
expense of the peripheral regions retaining only traditional activities.
• At the intermediate level, within particular countries, the existence of cities is
treated as evidence of the role played by increasing returnsto scale. The new
economic geography examines whether the changes in the interaction of cen-
tripetal and centrifugal forces will tend to promote the development of met-
ropolitan areas or foster their fragmentation.
• Small scale agglomerations are also observed through the existence of tech-
nological or industrial districts (Silicon Valley, the City in London, the North
East Central Italy or “Third Italy”). These agglomerations result from a par-
ticular process of cumulative causation linked to the nature of development.
In consequence, the location of particular industries often reflects the lock-in
of transitory advantages.

The idea behind this working paper is to give some elements for an economic
interpretation of urban phenomenon, referring to the main results of the New
Economic Geography. Section II of the paper presents the main concepts used in
the theoretical approach of the new economic geography. Section III describes
the standard Core-Periphery model and its extensions. Section IV focuses on the
growth of cities, discussing the complementarities between growth and location
theories and looking at the role of information and communication technologies
on the determinants of location decisions by firms.

4
 Working Paper 16-02

II The New Economic Geography:

background and main features

A. The resurgence of geography

In the 1990s, the issues of space and location started to get increasing attention in
the economic literature, aiming at explaining the spatial concentration of eco-
nomic activity and regional growth disparities. This research was mainly initi-
ated by the work of Krugman (1991a, 1991b), who contributed to the
development of a general equilibrium approach to study geographic agglomera-
tion, industrial clustering and the evolution of cities.

The framework proposed by the new economic geography (NEG) integrates vari-
ous elements derived from traditional theory about the location of economic
activity1. However, the traditional theory said very little about the causes of
agglomeration. The new economic geography proposes general equilibrium
models in which the spatial distribution of economic activities can be explained
by endogenous location decisions. The interactions between the different mar-
kets, between the firms and their suppliers and customers, the role of workers as
production factors and consumers are the key elements of these models.

Regional economics and urban economics have already analysed the potential
advantages from geographical concentration of economic activity. The centripe-
tal forces favouring agglomeration of firms within industries have already been
put forward by Marshall (1920) who distinguishes the advantages of having a
large local market, a large labour pool and knowledge spillovers.

Harris (1954) and Pred (1966) studied the emergence of large regional concentra-
tion of economic activity, such as the “manufacturing belt” in America's northeast
and inner Midwest. Harris (1954) emphasized the role of access to markets in the
location of economic activities. He measured the market access of each region us-
ing a measure of “market potential” defined as a weighted sum of purchasing
power across locations, with the weights for each location depending inversely
on its distance. Harris concluded that the heavily industrialised regions of the
United States were in general also locations with exceptionally high market po-
tential. He also noted that the concentration of production was self-reinforcing.
Pred (1966) was interested in the dynamics of regional growth and in the condi-
tions for a regional economy to take off in a cumulative process of growth.

The advantages of concentration resulting from interactions between different

sectors and many of the underlying ideas about cumulative causation through

1. For an overview of traditional theory (e.g. von Thünen, Weber, Christaller, Lösch), see Fujita et
al. (1999).

5
Working Paper 16-02

forward and backward linkages were also familiar to development economics in

the 1950s (Myrdal, 1957; Hirschman, 1958).

B. Increasing returns to scale

The reason why mainstream economic theory rediscovered economic geography

only recently is mainly because increasing returns to scale, associated with indi-
visibilities in production, are crucial in the explanation of agglomerations.
Under constant returns to scale, firms can locate anywhere. It is the presence of
increasing returns that induce firms to concentrate operations in one location
rather than in several locations in order to work more efficiently.

Therefore, a distinctive feature of the new economic geography is that it makes

use of the analytical tools provided by industrial organisation theory and devel-
oped in the fields of international trade theory1 and economic growth theory2 in
order to model increasing returns and imperfect competition3. The new eco-
nomic geography draws heavily on a spatial Dixit-Stiglitz monopolistic competi-
tion model. The inclusion of transport costs and space requires a certain number
of modelling devices, as will be described in the next section.

The key elements of the NEG models are the explicit consideration of space due
to transport costs, the accounting for economies of scale, the microeconomic
foundation of centripetal and centrifugal forces and pecuniary externalities,
which are the endogenous outcome of market forces. Economies of scale and
transport costs cause imperfect competition, which is captured by the Dixit-
Stiglitz monopolistic competition approach.

The NEG considers that the geographic concentration of economic activity

reflects the interaction between the presence of increasing returns to scale and
transport costs. In such a world, increasing returns activities are predicted to
locate in the larger market, giving rise to a “home-market effect”4. When trans-
port costs matter, attractive locations for production of goods subject to econo-
mies of scale are those locations, which are close to markets (backward linkages)
and suppliers of intermediate goods (forward linkages), other things being
equal. Then, concentration of production in some location tends to attract the
mobile factors of production. Workers have better job and consumption oppor-
tunities where production is concentrated. The resulting concentration of the
labour force leads to more demand for consumption goods in that location,
which makes the region more attractive for producers. Some centrifugal forces
work against these centripetal forces. Concentration of population and economic
activities in one region may drive land rents and housing prices up and lead to
congestion and environmental problems. Moreover, as immobile factors remain
in peripheral areas, firms from the centre may want to move there to supply
these areas.

1. See Helpman and Krugman (1985) for a survey of the “New International Economics”.
2. The “New Growth Theory” goes back to the work of Romer (1986, 1990). See Barro and Sala-i-
Martin (1995) for a survey of the literature.
3. If increasing returns are external to firms (unit costs depend on the size of the industry), the tools
of competitive analysis can still be used. If increasing returns are internal to firms (unit costs
depend on the firm's size), the model has to introduce imperfect competition.
4. The home-market effect, predicted by models of trade based on increasing returns to scale, is the
tendency for large countries to be net exporters of goods with high transport costs and strong
scale economies (Hanson, 1998).

6
 Working Paper 16-02

C. Agglomeration economies

The new economic geography starts from the observation that economic activity
is unevenly distributed across space. The concentration of economic activity can
be only partly attributed to differences in underlying characteristics (geography,
technology, endowments). In fact, a main explanation of the spatial concentra-
tion of firms and consumers/workers is related to the existence of externalities,
the so-called 'agglomeration economies', which means that spatial concentration
itself creates a favourable environment for the location of economic activity
(Fujita et al., 1999), enhancing productivity and growth.

The new economic geography distinguishes between pecuniary and non-pecu-

niary externalities in reference to Scitovsky (1954), as mentioned by Fujita and
Thisse (2002).

Pecuniary externalities refer to the effects of interactions mediated by markets.

These interactions, which contribute to the formation of agglomeration econo-
mies, can be divided into two types:
- Interactions between firms and households: these externalities relate to
employment or consumption of goods and services. The proximity
between firms and households facilitates the matching process in terms of
skills on the labour market and the access to a larger variety of goods and
services.
- Interactions between firms: the proximity between firms facilitates for-
ward-backward inter-industry and intra-industry linkages such as the
access to a larger variety of intermediate goods and business services, as
well as intra-firm relations between front office and back office.

Non-pecuniary externalities, such as “knowledge spillovers” (or technological

externalities), refer to the effects of non-market interactions due to proximity.
Non-pecuniary externalities are realised through processes directly affecting the
utility of an individual or the production function of a firm. Non-pecuniary
externalities are thought to be an important factor in the creation of agglomera-
tions. But, by their nature, they are difficult to measure empirically. Such con-
cepts as informational and technological externalities between firms, or in
general informational exchanges between agents explain why households and
firms want to cluster together. The reason for clustering is the fact that these
externalities between firms are assumed to decline rapidly with distance.
Knowledge spillovers are channelled through face-to-face communication and
casual diffusion of information between firms. These non-market mechanisms
matter most for small scale agglomerations.

In the literature (Duranton, 1997; Fujita and Thisse, 1996), another distinction is
made between “location economies” associated to firms of a same sector (intra-
industry externalities) and “urbanisation economies” associated to firms located
in a same area (inter-industry externalities). Some authors have demonstrated
the positive impact of a diversified sectoral environment on the results in terms
of innovation (Audretsch and Feldman, 1999; Duranton and Puga, 2001).

7
Working Paper 16-02

D. Cumulative or circular causation mechanism

The new economic geography considers that the concentration of economic

activity in a location creates a favourable economic environment that supports
further concentration. The interaction of demand, increasing returns and trans-
port costs creates a circular causation process. In the presence of increasing
returns and transport costs, firms tend to agglomerate in a single place and to
choose a location with a large local demand. But the presence of more firms in a
single place then creates an incentive for other firms to locate in the same place.
The location decisions of firms and consumers/workers form a self-reinforcing
process.

The circular relationship, in which the location of demand determines the loca-
tion of production, and vice-versa, can be a deeply conservative force. Circular
causation reinforces small differences in the production structure and can differ-
entiate similar markets into large and small. Locations with large population
will tend to specialise in the production of goods for which scale economies,
product differentiation and transport costs are significant. With a large home
market, producers of highly differentiated products can potentially obtain
enough local demand to exploit economies of scale. Locations with small home
markets will tend to specialise in standard products, or products for which
transport cost or scale economies are insignificant.

Ottaviano and Puga (1998) distinguish three kinds of cumulative mechanisms

through which economic activities can agglomerate: (1) migration-induced
demand linkages, (2) input-output cost and demand linkages, (3) endogenous
growth, factor accumulation, and intertemporal linkages.

E. Centripetal and centrifugal forces

The spatial distribution of economic activity is very uneven. Many activities are
highly concentrated geographically, although not in a single location. This is a
consequence of the interaction between agglomeration forces and dispersion
forces formalised in economic geography models.

The centripetal (agglomeration) forces, i.e. forces that pull economic activity
towards existing locations of economic activity, have already been put forward
by Marshall (1890). Following the exposition of Marshall's ideas in Krugman
(1998b), and in Fujita and Thisse (2002), the firm faces three potential sources of
centripetal forces:
- Backward and forward linkages: the large size of the local market creates
both backward linkages (large markets are preferred locations for produc-
tion of goods subject to economies of scale) and forward linkages (large
markets support the local production of intermediate goods, lowering
costs for downstream production).
- Pooling of skilled workers: the concentration of economic activity favours
the availability of specialised labour skills.
- Externalities: the concentration of economic activity creates externalities,
such as “knowledge spillovers” (technological spillovers), i.e. an improve-
ment of the diffusion of information due to proximity (face-to-face con-
tacts).

8
 Working Paper 16-02

Empirical studies tend to show that the effects of knowledge spillovers and the
availability of specialised labour skills are more relevant for explaining small
scale agglomeration (e.g. Silicon Valley, the City of London), while the effects of
forward and backward linkages are more relevant for large scale agglomeration
(e.g. pattern of location in Europe).

Besides agglomeration forces, the presence of dispersion forces explains the

existence of various concentration places. The centrifugal forces, i.e. forces that
push economic activity away from existing centres, are due to the existence of
transport costs in the face of immobile factors (land, natural resources, agricul-
tural workers and consumers), and negative externalities. In a survey of the liter-
ature, Ottaviano and Puga (1998) mention models with congestion externalities,
according to which high concentration produces a rise in the price of immobile
local factors (land rents and wages) and goods, which can induce dispersion
effects.

Competion between firms can also represent a dispersion force (Duranton,

1997). There is a trade off between a location in proximity of consumers at the
price of a higher degree of competition and a location more distant from the cen-
tre in order to benefit from a lesser degree of competition, at the cost of losing
access to a large concentrated market

TABLE 1 - Forces affecting geographical concentration

Centripetal forces Centrifugal forces

Market-size effects (backward-forward linkages) Immobile factors
Thick labour market (availability of specialised labour skills) Land rents
Pure external economies Pure external diseconomies

Source: Krugman (1998b).

In the real world, all these centripetal and centrifugal forces affecting geographi-
cal concentration are at work. Theories, however, tend to focus on a limited set of
forces. If the aim of empirical work is to understand better why economic activ-
ity tends to concentrate geographically, a distinction has to be made between
competing theories that highlight the different mechanisms through which scale
economies contribute to agglomeration.

The simple model developed in Krugman (1991a, 1991b) suggests that agglom-
eration is the result of demand linkages between firms, which are created by the
interaction of transport costs and fixed costs in production. Scale economies are
internal to the firm. The presence of immobile factors acts as an opposite disper-
sion force. Other models introduce scale economies in producing non-traded
intermediate inputs (Rivera-Batiz, 1988) or vertical stages of production, in
which firms produce both consumer and industrial goods (Venables, 1996). In
these models, cost and demand linkages between industries (inter-industry
interactions) foster agglomeration. Alternative models suggest that agglomera-
tion economies are the result of positive spillovers between firms in the same
location (intra-industry interactions), i.e. location-specific externalities (Hender-
son, 1974). Scale economies are external to the firm. The source of these intra-
industry externalities is not specified. However, following Marshall (1920),
agglomeration occurs because the proximity of firms in one same industry facili-
tates learning and the exchange of ideas.

9
Working Paper 16-02

In a dynamic model of city formation, Black and Henderson (1999) combine

location-specific industry externalities with location-specific human-capital
externalities according to which the accumulation of human capital generates
positive spillovers (Lucas 1988), so that the concentration of labour in a given
location makes all local workers, regardless of industry, more productive. Con-
gestion costs associated with the limited local availability of housing or other
non-traded goods or factors act as a centrifugal force. As cities develop, the price
of housing increases in urban areas relative to peripheral areas. To attract work-
ers to cities, firms must compensate them for the relatively high costs of urban
living and commuting. The higher productivity of labour in agglomerated
regions justifies these higher wages. If agglomeration economies are sufficiently
strong, most production occurs in industry clusters where wages and housing
prices are relatively high.

10
 Working Paper 16-02

III The Core-Periphery model

Economic geography models rely on the interaction of increasing returns to scale

(imperfect competition), transport costs, and factors mobility. The NEG models
use a common set of tools involving the Dixit-Stiglitz monopolistic competition
framework and the so-called “iceberg” transport costs (see infra). These particu-
lar tools are considered the best available for dealing with increasing returns,
transport costs and backward-forward linkages in an analytically tractable gen-
eral equilibrium framework.

A. The spatial impossibility theorem

The spatial impossibility theorem of Starrett (1978) tells that: if space is homoge-
nous and transport is costly, there does not exist any competitive equilibrium
with trade between distant locations (Ottaviano and Thisse, 2001). A model with
mobile agents on a closed, homogeneous space, facing a production technology
with constant returns to scale (no indivisibility), can never explain the occur-
rence of agglomerations. In such a framework, economic activity will disperse
without any countervailing force, because dividing up production over many
locations leads to no loss in efficiency. With the removal of constant returns to
scale, the assumption of imperfect competition is needed to capture the main
source of agglomeration: indivisibility.

B. The modelling tools

1. Dixit-Stiglitz monopolistic competition framework

The monopolistic competition model presented by Dixit and Stiglitz (1977) can
be used to circumvent the problems deriving from the introduction of market
structures consistent with the presence of increasing returns to scale. This model,
known for its applications to a new class of growth and trade theories, is based
on a convenient assumption on market structure (monopolistic competition) to
avoid the problems due to price-taking behaviour in the presence of increasing
returns to scale. The model also disregards the complex mechanisms of strategic
interactions, as firms ignore the effects of their actions on prices and income.

The Dixit-Stiglitz model uses specific functional forms for consumer preferences
allowing for a 'preference for variety'. This means that the utility of a consumer
will be positively related to the number of goods available. To completely elimi-
nate every producer's market power, it is often assumed that the range of goods

11
Working Paper 16-02

is continuous, and each producer is infinitely small. With full competition, the
good is assumed to be homogeneous, and its price the only criterion of selection.
With monopolistic competition, consumers discern different varieties, and prod-
ucts from different producers are imperfect substitutes. Even if each individual
producer faces increasing returns to scale in production, the largest producer is
not always able to eliminate smaller competitors because substitution between
products is limited.

According to Quigley (1998), the NEG models based on the Dixit-Stiglitz

approach can be particularly useful in the context of urban economics, to explain
why diversity in consumer goods and in production inputs can produce external
scale economies. The intuition behind this result is that the size of the city and its
labour market will determine the number of specialised local goods and pro-
ducer inputs given the degree of substitutability among these goods and among
these inputs. A larger city will have a greater variety of these goods and inputs
and this greater variety increases utility and output. On the consumption side,
greater differentiation among goods means that variety has greater effect on util-
ity. On the production side of the economy, in a similar way, output will be
related to the number of inputs available. Greater differentiation among inputs
means that variety has greater effect on output. Thus, larger cities will be offer a
broader range of products and the well-being of their inhabitants will increase
with size, up to a certain limit.

2. Transport costs

Economic geography models involve the assumption of iceberg-type transport

costs. This assumption first introduced by Samuelson (1954) in international
trade means that a fraction of any good shipped simply “melts away” in transit -
usually at a constant rate per distance covered. So, transport costs are incurred in
the good itself and one needs not model the price of transport separately.

C. The standard Core-Periphery model1

The basic model provided by Krugman (1991a, 1991b) combines a monopolistic

competition model and pecuniary externalities associated with forward-back-
ward linkages to show how large-scale agglomerations can emerge. Fundamen-
tal for the emergence of industrial concentration are economies of scale due to
which the profit maximising manufacturing firms only locate in one region.
Under certain assumptions backward and forward linkages promote a self-rein-
forcing industrial concentration process. The driving force behind this is the
mobility of workers and firms, which locate where the market is relatively large.
The immobility of the farmers acts in the opposite direction. Therefore, the spa-
tial equilibrium structure, which results from the relative weight of the centripe-
tal and centrifugal forces, crucially depends on the level of the transport costs,
the manufactures' share of total expenditure and the elasticity of substitution
among the differentiated manufacturing goods.

1. The following model is the version of the core-periphery model of New Economic Geography presented in Fujita M., Krugman P.,
Venables A. (1999). For convenience the same notation as in the book is used.

12
 Working Paper 16-02

1. Assumptions

The model considers two regions and two sectors. Agriculture tied to land is
perfectly competitive and produces a homogeneous good under constant
returns to scale. The total quantity is CA. Manufacturing that can be located in
either region is monopolistically competitive and produces differentiated prod-
ucts (i = 1,…,I) under increasing returns to scale.

Aggregate utility from consuming both types of goods is a Cobb-Douglas func-

tion of agricultural product and of a CES sub-utility function derived from con-
suming manufactures.

µ 1–µ
(1) U ( C A ,C M ) = CM CA
where µ ∈ [ 0 ,1 ]

Let “ci” be the quantity consumed of manufactured product i. The utility of the
consumer derived from (c1,…,cI) is given by a CES function that captures symme-
try and preference for variety.

σ -
I σ – 1- σ -----------
----------- –1
σ
(2) C M = ∑ ci
i=1

where σ>1
With this specification of utility, it results from the household's maximisation
problem that the share of consumer expenditures devoted to manufacturing
goods in household equilibrium is µ and the elasticity of substitution between
different manufactured products is constant (σ).

There is a single production factor in the economy, but each sector uses a specific
type of labour: farmers in agriculture, workers in manufacturing. The supply of
labour is given exogenously. To simplify, the model assumes that the share of
manufacturing workers in the population equals µ, the share of manufacturing
in consumer expenditures.

Consumers maximize their utility given their budget constraint. There is free
entry for firms and firms maximize profits.

2. Conditions for agglomeration

The economy consists of two regions. The scope of the model is to show how
manufacturing is distributed across regions and under what conditions the
entire manufacturing population will concentrate in one region.

13
Working Paper 16-02

Krugman supposes first that both types of labour are immobile, so that the dis-
tribution of workers and farmers across regions is fixed. The transport costs for
manufactured goods between regions are introduced in the model by iceberg
transport costs. This means that of any unit of the good transported from a
region to the other one, only 1-τ units arrive at the destination, with the parame-
ter τ a positive constant between 0 and 1. The model assumes that transport costs
for agricultural goods are non-existent.

Several intermediate results, given this set-up, are worth noting (Schmutzler,
1999). First, producers are usually assumed to face a fixed cost “F” and a variable
cost “a” per item produced. As the fixed cost per product declines with total pro-
duction, they are subject to an increasing returns technology. However, because
of the downward sloping consumer demand, output cannot grow indefinitely.
Instead, producers maximize profits by setting marginal benefit equal to “a”.
Facing demand generated by the utility function (1), this pricing strategy results
in a constant mark-up over marginal costs. Second, because of increasing returns
to scale, each firm produces only one product. Third, free entry sets pure profits
equal to zero. Fourth, each firm produces at the same output level in equilib-
rium.

The equilibrium output of each firm is a positive function of the fixed costs and
the elasticity of substitution, and a negative function of the marginal costs. The
number of firms in a region is a positive function of its manufacturing labour
supply, and a negative function of both fixed and marginal costs.

Then, Krugman introduces labour mobility in the model and analyses the con-
frontation of centripetal and centrifugal forces. While agricultural labour contin-
ues to be immobile, manufacturing population moves towards the region that
offers the higher present real wage. In equilibrium, both regions must either
offer the same manufacturing wage or the manufacturing population will be
concentrated in the region offering the higher real wage. Such a concentration of
the entire manufacturing activity in one location in equilibrium can only occur if
no firm has an incentive to locate at the periphery. With transport costs, it is con-
venient for workers to stay in the centre because the cost of living is higher in the
periphery, since most consumption goods have to be imported from the centre. It
is also cheaper for firms to serve from the centre the majority of customers living
in the centre. But, it is cheaper to serve the agricultural population in the periph-
ery if firms produce in the periphery. An agglomeration equilibrium results
when this latter centrifugal effect is small relative to the other two centripetal
effects.

In this model, agglomeration depends on the level of transport costs, the size of
the manufacturing sector and the consumer preference for variety. Reasonably
high transport costs are a condition for agglomeration, so that serving the
periphery from the centre is a feasible alternative to local production. The size of
the manufacturing sector reinforces agglomeration since the wage premium nec-
essary to compensate workers in the periphery will increase, since a large quan-
tity of goods has to be delivered from the centre. The preference for variety also
matters, because a small elasticity of substitution goes along with high econo-
mies of scale, which favours firms’ location in the centre.

14
 Working Paper 16-02

An interesting feature of the model is that minor initial differences between

regions may lead to a core-periphery structure. So “history matters” (Arthur,
1989). Supposing that region 1 has a slightly larger share of agricultural popula-
tion, Krugman investigates how transport costs influence the distribution of
manufacturing between the two regions. With relatively high transport costs, an
equilibrium will emerge such that both regions have some manufacturing, but
region 1 has a higher manufacturing share than region 2. The larger market is
more attractive. As transport costs come down it is worth moving to the larger
market to ship products out to region 2. At some point, a new equilibrium
emerges where all manufacturing is concentrated in region 1. The advantages of
concentrating production dominate the advantage of being close to the market.

D. The extentions of the Core-Periphery model

The core model developed by Krugman (1991a, 1991b) is in general not suited
for empirical validation. The main reason is that for an intermediate range of
trade costs it produces only one location with manufacturing economic activity.
This result does not correspond, obviously, to the facts about the spatial distribu-
tion of manufacturing activity in the US or in Europe. Moreover, the core model
lacks some of the spatial characteristics of agglomerations, which have been
found to be very relevant empirically, such as the tendency of prices of local
goods to be higher in agglomerations.

Krugman and other authors modified the prototype NEG model in various
respects. In some model extensions, different factors in favour of industrial dis-
persion like non-tradable goods and congestion costs are introduced. But also
elements reinforcing industrial clustering, such as vertical linkages between
industries, are considered. But neither the fundamental structure of the basic
model nor its key mechanisms are modified by those extensions. A common fea-
ture of most of the model extensions is that they do not deliver analytical solu-
tions and exhibit multiple equilibria. In general, numerical examples illustrate
the model mechanisms. They contribute to support that the new economic geog-
raphy provides general and robust insights about the interactions of various
forces and their impact on the geographical distribution of economic activities.

Krugman and Venables (1995) present a model in which they assume, in contrast
with Krugman (1991a, 1991b), no inter-regional labour mobility. In consequence,
when a sector expands the labour supply must come from other sectors in that
region. Cumulative causation and the emergence of agglomeration in this model
come from input-output linkages between firms, which are now assumed to use
each others’ outputs as its inputs. In this model, firms benefit from being close to
each other by avoiding transport costs on factors of production. This model pro-
duces two types of equilibria. For high trade costs of manufactures, a symmetric
equilibrium, and for low trade costs a core-periphery solution. For intermediate
transport costs, asymmetric but unstable equilibria are possible (Fujita, Krug-
man and Venables, 1999).

Krugman and Venables (1996) extend this model by assuming two manufactur-
ing sectors, each of which trades more with firms in the same sector than with
firms of the other sectors. Complete agglomeration is less likely, because
reduced costs and demand linkages benefit to firms in the same sector while

15
Working Paper 16-02

competition in product and labour markets applies to all firms in all sectors
equally. For low trade costs, this leads regions to specialise in one sector only.

An extension of this Krugman-Venables model consists in introducing an addi-

tional dispersion force into the model, assuming increasing-returns to scale in
agriculture (Puga, 1999; Fujita, Krugman and Venables, 1999). In this model,
agglomeration raises wages in the core region, making it attractive for firms to
relocate to a peripheral region where labour costs are lower. The result is a W-
type relationship between the share of industry in each region and trade costs.
For high trade costs, there is equal distribution of industrial activity, for interme-
diate levels of trade costs full as well as partial agglomeration results, and for
low trade costs there is a return to equal distribution of industrial activity.

Given the observation that full agglomeration is not in accordance with the facts,
geographical economics models based on forward and backward linkages and
with no interregional labor mobility (Venables, 1996; Krugman and Venables,
1996; Puga, 1999) seem therefore useful models for empirical testing. Unfortu-
nately, direct testing of this class of NEG models is rather problematic because it
requires detailed information on input-output linkages between firms on a
regional level. The model developed by Helpman (1998), with its empirical
applications by Hanson (1998, 1999) is a useful alternative for empirical research.
It combines elements of the core model such as demand linkages with the intro-
duction of a non-tradable consumption good (i.e housing). The price of housing,
which increases with agglomeration, acts as a dispersion force in this model, in
the same way as the rising wages in the previous model (Puga, 1999). In fact, it is
shown that both models produce similar results in terms of equilibrium out-
comes.

16
 Working Paper 16-02

IV Cities and economic growth

The share of population living in cities has tended to increase over the long
period. Predicting the evolution of cities requires a framework for understand-
ing the costs and the benefits of urban areas. This section examines what can be
said about the agglomeration forces and the dispersion forces in order to assess
how changes in these forces will affect the growth of cities.

A. Economic growth and agglomeration

According to the neoclassical growth theory, economic growth was understood

as the accumulation of physical capital (Solow, 1956). The inequality between
regions was explained by an unequal distribution of physical capital. This view
has been challenged by the endogenous growth theory, where total factor pro-
ductivity (TFP) is considered the engine of economic growth. These new theories
relate growth to increasing returns and to technological progress (e.g. Romer,
1986 and 1990; Grossman and Helpman, 1991; Aghion and Howitt, 1998).

Aggregate output Y can be derived by the following production function (Quah,

2001):

(1) Y = F ( K, AN )
where F is increasing and concave, K denotes the stock of physical capital, N
denotes the quantity of labour employed, and A the residual factor.

In the endogenous growth theory, A can be interpreted as the level of technol-

ogy. It is the total factor productivity (TFP)1.

Writing y = Y/N for labour productivity or output per worker and k = K/N for
the capital-worker ratio, and assuming that F satisfies constant returns to scale,
equation (1) implies:

·
(2) y· ⁄ y = ( 1 – s K ) ( A· ⁄ A ) + s K ( k ⁄ k ) ,

FK ∂F-
where s K = K  ------ , and F K = ------
F ∂K

1. This view is opposed to the Solow interpretation of A as a mere residual from econometrically
fitting the production function, given Y, K, and N.

17
Working Paper 16-02

With competitive markets, the term sK is physical capital’s income share. Equa-
tion (2) decomposes growth in output per worker into contributions due to tech-
nology and to physical capital. While attention first focused on the contribution
of physical capital (second term on the right of (2)), more recent research (Lucas,
1988; Romer, 1994) suggests that the A term is crucial for understanding eco-
nomic growth.

The new economic growth theories suggest that TFP reflects increasing returns
and/or technological progress, i.e. the outcomes of science and research and
development (e.g. Aghion and Howitt, 1998; Grossman and Helpman, 1991;
Romer, 1986, 1990). Both explanations can be distinguished. However, most
researchers follow Romer (1990) and suppose that technological change evolves
endogenously due to research and development, an economic activity that
involves increasing returns (Quah, 2001). With increasing returns, economic
activity can display technological lock-in (David, 1985; Arthur, 1989). Thus, once
a technology establishes a benchmark, it tends to last even if preferable alterna-
tive technologies become available, because the economic agents find it unprofit-
able to switch to other technologies.

Interestingly, this reasoning is similar to that followed in the new economic

geography, as explained in the previous sections. Krugman (1991a and 1991b)
emphasizes the role of increasing returns to explain regional specialisation in
specific industries, the evolution of cities and geographical agglomeration. His-
torical accident can bring a specific industry to a particular location. Then, this
industry will find it advantageous to cluster there due to increasing returns to
scale and to forces such as technological lock-in.

B. Growth in cities

Some theories of city growth stress the role of dynamic externalities, particularly
knowledge spillovers (Glaeser et al., 1992; Black and Henderson, 1999b), there-
fore converging with the new theory of economic growth, which considers exter-
nalities as the engine of growth. Such models argue that the concentration of
firms and workers in cities provide an environment in which the free flow of
ideas is facilitated. The interactions between people in cities encourage learning
and innovation, which create externalities for firms. This occurs because innova-
tions and improvements in one firm improve the productivity of other firms
without full compensation. Geographical proximity is a condition of face-to-face
communication, which makes externalities particularly large in cities. Therefore,
knowledge spillovers might explain why cities survive despite high rents.

1. Knowledge spillovers

Various theories have been proposed to explain the role of knowledge spillovers
for city growth. They differ in whether knowledge spillovers come from within
the same industry or from other industries, and how local competition affects
these externalities with respect to growth.

The first hypothesis, originally developed by Marshall (1890) and later formal-
ized by Arrow (1962) and Romer (1986) (MAR), emphasizes spillovers between

18
 Working Paper 16-02

firms in the same industry. This means that the concentration of an industry in a
city facilitates knowledge spillovers that arise, for example, through the dissem-
ination of ideas and the movement of skilled labour among neighbouring firms.
These spillovers are thought to be most important when local competition is
weak. Innovation and growth are encouraged when rents associated with sector-
specific knowledge can be internalised.

The second hypothesis, developed by Porter (1990), also argues that knowledge
spillovers in a geographically concentrated industry are most important, but in-
sists that their effects on growth are enhanced by strong local competition, which
forces firms to innovate in order to survive.

The third hypothesis, developed by Jacobs (1969), emphasizes knowledge spillo-

vers across industries, occurring through the “cross-fertilization of ideas”.
Diversity of geographically agglomerated industries promotes innovation and
growth in cities. Like Porter, Jacobs supports the view that these externalities are
most effective in an environment of strong local competition.

TABLE 2 - Factors favouring knowledge spillovers

Theories Specialisation vs diversification Monopoly vs local competition

MAR Specialisation Monopoly
Porter Specialisation Local competition
Jacobs Diversification Local competition

2. Agglomeration forces and congestion forces in cities

The attractiveness of cities can be seen in the quantity of people living in urban
areas. It is also reflected by the wage premium paid to workers in those areas.
Higher wages in cities can indicate a higher productivity of workers, but this
should not underestimate the benefits created by agglomeration.

According to Glaeser (1998), several factors explain why firms are willing to pay
to locate in cities. These factors are linked to reduced transport costs for goods,
people and ideas.
• Transport costs and increasing returns of some sort play a major role in urban
and regional economics to explain cities and locational choices. But while
transport costs for goods continue to matter, they have become much less
important. There is the relative decline of traditional manufacturing charac-
terised by fixed setup costs and high transport costs and the corresponding
rise of services. Transport costs have also declined.
• Another important benefit of cities comes from the elimination of distance
between people. The advantage of labour market pooling is to facilitate the
access of producers to skilled workers, to improve the division of labour, and
to insure workers against firm- or industry-specific shocks.
• The geographical proximity can also enhance the diffusion of ideas across
firms by the movement of workers across firms and the exchange of ideas in
both formal and informal settings. Urban density also encourages learning

19
Working Paper 16-02

through the interactions between people, which can accelerate human capital
accumulation.

While cities are growing, the benefits of agglomeration can, at some point, be
overwhelmed by the congestion effects. They include the costs of living, pollution
and social problems.
• The costs of living include housing and commuting costs. There is no ques-
tion that houses in cities cost more and there is a correlation between com-
muting costs and the city size. These costs are likely to remain despite
improvements in building technology and transports.
• The evolution of pollution costs associated with cities could be tempered
through better emissions controls, changes in transport technology and the
decline of manufacturing industry in cities.
• Cities attract the poor because they have a better access to transport infra-
structure and a series of public goods. Basic urban theory implies that the
rich leave cities because they have a greater demand for space, which is
cheaper outside of cities. Government action is therefore needed to encour-
age diversified urban areas and attenuate social problems in cities.

3. Optimal degree of concentration

There may well be an optimal degree of concentration, according to Henderson

(2000), which should neither be too small relative to the efficient size to reap the
benefits from agglomeration, nor too large to avoid the problems of large cities
such as high commuting, congestion and living costs, and low quality of urban
services.

This optimal degree of concentration is also linked to the degree of economic

development. At an early stage of economic development, it is rather efficient to
have spacially concentrated infrastructure in order to enhance information spill-
overs and knowledge accumulation. As development proceeds, a greater disper-
sion of economic activity eventually becomes efficient, because it is possible to
spread infrastructure and knowledge resources to the periphery and because
congestion locations are less efficient for producers and consumers. Deconcen-
tration occurs by manufacturing moving first from the core cities to nearby cities
and then to hinterland cities, where wage and land costs are lower.

Oversized cities, which lead to efficiency losses and a reduction of the urban
quality of life, can result from imperfectly working land markets and social mar-
ginal costs of increasing city size exceeding marginal benefits. Political institu-
tions can also encourage oversized cities. This result comes from the tendency of
national governement to invest less in inter-regional infrastructure, so that firms
are encouraged to locate in large cities, especially in the national capitals, rather
than in the hinterland.

20
 Working Paper 16-02

C. ICT in growth and agglomeration

1. Some features of ICT

According to Quah (2001), ICT significantly displays the same features that
induce both growth and agglomeration, i.e. increasing returns and knowledge
spillovers. On the one hand, ICT is an example of extensive technological
progress and represents a factor contributing directly to economic growth. On
the other hand, ICT is mainly an industry with increasing returns1. Therefore,
agglomeration and lock-in will characterise ICT while, at the same time, ICT con-
tributes to overall economic growth.

However, ICT has also two important specificities. First, various ICT products are
not affected by distance. Their trade implies no transport costs. So, why should
2
ICT industry concentrate geographically. ICT clusters have to be explained . A
related question is whether the co-location of diverse industries or the specialisa-
tion on a narrow set of activities has the greatest impact on innovation and
growth. Second, ICT products are non rival, a characteristic generally displayed
by intangible assets. But, ICT products are widely used by consumers directly
and they are protected by copyright rather than patents.

2. “The death of distance”

A frequently asked question is about the impact of ICT on the balance between
centripetal and centrifugal forces underlying the location of economic activities.
Some agglomeration forces are to be weakened by ICT, while other agglomeration
forces are likely to be unaffected by new technology.

New technologies will have a mixed effect on the cost of distance (Venables,
2001). transport costs via internet are zero. But, only some activities that can be
codified and digitised will be appropriate to distance supply. Thus, some activi-
ties will no longer need to be close to consumers and will be in search of lower
cost locations. Airline ticketing services, back-room operations of banks and call
centres are well-known examples of such activities. As internet and telecommu-
nications infrastructures expand, workers will no longer have to be together to
communicate. However, communication in specific matters is better achieved
face-to-face. In consequence, knowledge spillovers that firms derive from prox-
imity to other firms will remain important. For many activities, the access to
skilled labour and to local networks of specialised suppliers will continue to ex-
plain the propensity of economic activity to cluster. The persistence of locations
such as Silicon Valley3 reflects not just the specificities of ICT, but the high-skilled
nature of that industry.

1. Quah (2001) distinguishes different parts in the ICT industry and he is well aware that increasing
returns are more obvious in certain parts than in others. For example, in the computer industry
including software, hardware and services, it is in the software part that increasing returns are
mostly expected.
2. The distinction between ICT industry and ICT products can be part of the explanation. While
many ICT products may no longer be affected by distance, it is not necessarily the case for ICT
industry.
3. The activities located in the Silicon Valley are intensive in research and development.

21
Working Paper 16-02

Therefore, the economic landscape will become only progressively less agglom-
erated as a consequence of the introduction of ICT. Some activities, which are
intensive in knowledge, such as research and development, and requiring face-
to-face communication will remain geographically concentrated. Other activities
less dependent on this kind of communication will be able to relocate to a small
number of places.

22
 Working Paper 16-02

V Conclusion

The main contribution of the “new economic geography” is to explain the cumu-
lative causation process leading to agglomeration in a theoretical framework with
solid micro foundations, thereby giving behavioural underpinnings to the in-
sights of previous location theory and regional science. However, no model
focusing on a single cause can really capture the complexities of any problem.
While Krugman (1991a,1991b) emphasises the role of pecuniary externalities, al-
ternative theories explaining the relevance of location in various issues contest
this option. Empirical studies tend to show that knowledge spillovers and the
availability of specialised labour skills may well prove to be the determining forc-
es behind small scale agglomeration (Audretsch and Feldman, 1996), while
forward-backward linkages are more relevant for patterns of regional location in
Europe.

A common objection to the new economic geography is that mathematical for-

malism leads to a number of simplifying assumptions. In fact, while economic
geographers prefer to build bottom-up explanations, most economists prefer us-
ing explicit models. Krugman argues that the mechanisms described in the new
economic geography are generally consistent with empirical observation. But, ac-
cording to Neary (2001), no serious attempts are made to compare explanations
related to agglomeration with other possible explanations. Moreover, omitting
strategic considerations, the new economic geography models based on the Dixit-
Stiglitz monopolistic competition approach miss an important aspect of reality
developed in games theory.

The empirical research is still at an early stage to draw policy conclusion. How-
ever, the availability of new data sets has encouraged the development of a
growing body of empirical literature directly based on new economic geography.
In Europe, the new economic geography coincided with the debate on the effects
of EU deepening through greater economic and monetary integration and the ef-
fects of EU widening through enlargement. Thus, the new economic geography
served particularly as a conceptual framework for European regional policies
since it offered explanations as to why regional integration, by reducing transac-
tion costs, could lead to self-sustaining inequality. However, with its emphasis on
the positive effects of local spillovers and on economies of scale, this framework
also implies that there are positive effects from agglomeration and hence from re-
gional differencies. Martin (1999) shows the existence of a trade-off between
equity and efficiency at the spatial level. If economies of scale and localised spill-
overs explain the self-sustaining agglomeration process, then agglomeration
must have some positive efficiency effects. And, because infrastructure financed
by regional policies have an impact on transaction costs and therefore on the lo-
cation decision of firms, the long-term effect of certain regional policies may be
sub-optimal in terms of efficiency and growth.

23
Working Paper 16-02

There is also empirical research about the relation between urban concentration
and economic growth (Henderson, 2000). This literature shows that urban con-
centration in many countries appears to be excessive, which is an obstacle to
economic growth. It argues that investment in inter-regional transport infrastruc-
ture is a key policy instrument to reduce concentration and promote economic
growth.

24
 Working Paper 16-02

Bibliographie

Aghion Ph. and Howitt P. (1998), Endogenous Growth Theory, The MIT Press,
Cambridge.

Arthur W.B. (1989), “Competing technologies, increasing returns and lock-in by histor-
ical events”, Economic Journal 99, 116-131.

Arthur W.B. (1994), Increasing Returns and Path Dependence in the Economy, Ann
Arbor, The University of Michigan Press.

Audretsch D.B. and Feldman M.P. (1996), “R&D spillovers and the geography of in-
novation and production”, American Economic Review 86 (3), pp. 630-40.

Audretsch D.B. and Feldman M.P. (1999), “Innovation in cities: Science-based diver-
sity, specialization and localized competition”, European Economic Review 43,
pp. 409-429.

Barro R.J and Sala-i-Martin X. (1995), Economic Growth, McGraw-Hill, New-York.

Black D. and Henderson J.V. (1999a), “Spatial evolution of population and industry in
the United States”, American Economic Review 89, pp. 321-27.

Black D. and Henderson J.V. (1999b), “A theory of urban growth”, Journal of Politi-
cal Economy 107, pp. 252-284.

Braunerhjelm P., Faini R., Norman V., Ruane Fr., and Seabright P. (2000), Integra-
tion and the Regions of Europe: How the Right Policies Can Prevent Polarization,
Monitoring European Integration 10, Centre for Economic Policy Research,
London.

Ciccone A. and Hall R. (1995), “Productivity and density of economic activity”, Amer-
ican Economic Review, 86, pp. 54-70.

Ciccone A. (2002), “Agglomeration effects in Europe”, European Economic Review

46, pp. 213-227.

David P.A. (1985), “Clio and the economics of QWERTY”, American Economic Re-
view, 75, 332-337.

25
Working Paper 16-02

Dixit A.K. and Stiglitz J.E. (1977), “Monopolistic competition and optimum product di-
versity”, American Economic Review 67 (3), pp. 297-308.

Duranton G. (1997), “La nouvelle économie géographique: agglomération et disper-

sion”, Economie et Prévision, n°131, pp. 1-24.

Ellison G. and Glaeser E.L. (1999), “The geographic concentration of industry: Does
natural advantage explain agglomeration”, American Economic Review 89, pp.
311-16.

Fujita M., Krugman P., and Venables A. (1999), The Spatial Economy: Cities, Regions,
and International Trade, The MIT Press, Cambridge.

Fujita M. and Thisse J.F. (1996), “Economics of Agglomeration”, Journal of the Japa-
nese and International Economics 10, pp. 339-379.

Fujita M. and Thisse, J.-F. (2002), Economics of Agglomeration. Cities, Industrial Lo-
cation, and Regional Growth, Cambridge University Press, Cambridge and New
York.

Glaeser E.L. (1999). “Learning in cities”, Journal of Urban Economics 46, pp. 254-
277.

Glaeser E.L., Kallal H.D., Scheinkman J.A., and Shleifer A. (1992), “Growth in cit-
ies”, Journal of Political Economy 100, pp. 1126-1152.

Grossman G. and Helpman E. (1991), Innovation and Growth in the World Economy,
The MIT Press, Cambridge.

Hanson G. H. (1998), “Market potential, increasing returns, and geographic concentra-

tion”, Working Paper No. 6429, National Bureau of Economic Research.

Hanson G. H. (2001), “Scale economies and the geographic concentration of industry”,

Journal of Economic Geography 1, pp. 255-76.

Harris C. (1954), “The market as a factor on the localization of industry in the United
States”, Annals of the Association of American Geographers, 64, pp. 640-656.

Helpman E. and Krugman P. (1985), Market Structure and Foreign Trade, The MIT
Press, Cambridge.

Henderson J.V. (1997), “Externalities and industrial development”, Journal of Urban

Economics 42, 449-71.

Henderson J.V. (2000), “How Urban Concentration Affects Economic Growth”, Policy
Research Working Paper No.2326, World Bank.

26
 Working Paper 16-02

Hirschman A. (1958), The Strategy of Economic Development, Yale University Press,

New Haven.

Huriot J.M. and Thisse J.F., eds, (2000), Economics of Cities. Theoretical Perspectives,
Cambridge University Press, Cambridge and New York.

Jacobs J. (1969), The Economy of Cities, Vintage, New York.

Krugman P. (1991a), Geography and Trade, The MIT Press, Cambridge, Mass.

Krugman P. (1991b), “Increasing returns and economic geography”, Journal of Politi-

cal Economy 99 (3), pp. 483-499.

Krugman P. (1991c), “History versus expectations”, Quarterly Journal of Economics,

106, pp. 651-667.

Krugman P. (1998a), “Space: The Final Frontier”, Journal of Economic Perspectives

12(2), pp. 161-174.

Krugman P. (1998b), “What's New About the New Economic Geography?”, Oxford
Review of Economic Policy 14(2), pp. 7-17.

Krugman P., Fujita M., and Venables A.J. (1999), The Spatial Economy: Cities, Re-
gions, and International Trade, The MIT Press, Cambridge.

Krugman P. and Venables, A. (1995), “Globalization and the inequality of nations”,

Quarterly Journal of Economics 110(4), pp. 857–879.

Lucas R.E. (1988), “On the mechanics of economic development”, Journal of Monetary
Economics 22, pp. 3-42.

Marshall A. (1890), Principles of Economics, Macmillan, London.

Martin Ph. (1998), “Can regional policies affect growth and geography in Europe?”, The
World Economy 21 (6), pp. 757-774.

Martin Ph. (1999), “Are European regional policies delivering?”, EIB Paper, 4 (2), pp.
10-23.

Martin Ph. and Ottaviano G.I.P. (1999), “Growing locations: industry location in a
model of endogenous growth”, European Economic Review 43(2), pp. 281-302.

Myrdal G. (1957), Economic Theory and Under-developed Regions, London:

Duckworth.

Neary P. (2001), “Of hype and hyperbolas: Introducing the New Economic Geography”,
Journal of Economic Litterature 39, pp. 536-561, June.

27
Working Paper 16-02

Ottaviano G.I.P. and Puga D. (1998), “Agglomeration in the global economy: A survey
of the ‘New Economic Geography’ ”, The World Economy 21(6), pp. 707-731.

Ottaviano G.I.P. and Thisse J-F (2001), “On economic geography in economic theory:
Increasing returns and pecuniary externalities”, Journal of Economic Geography,
1(2), pp. 153-180.

Porter M. (1990), The Competitive Advantage of Nations, Free Press, New York.

Pred A.R.(1966), The Spatial Dynamics of US Urban-industrial Growth, 1800-1914,

MIT Press, Cambridge.

Puga D. (1999), “The rise and fall of regional inequalities”, European Economic Re-
view, 43(2), pp. 303-334.

Puga D. (2001), “European regional policies in light of recent location theories”, CEPR
Discussion Paper, no. 2767, London.

Quah D. (2000), “Internet cluster emergence”, European Economic Review 44, 1032-
1044.

Quah D. (2001), “ICT clusters in development: Theory and evidence”, Mimeo.

Quigley J. (1998), “Urban diversity and economic growth”, Journal of Economic Per-
spectives, 12 (2), pp. 127-138.

Rivera-Batiz F. (1988), “Monopolistic competition, economies of scale, and agglomera-

tion economies in consumption and production”, Regional Science and Urban
Economics, 18 (1).

Romer P. (1986), “Increasing returns and long-run growth”, Journal of Political Econ-
omy, 94, pp. 1002-1038.

Romer P. (1990), “Endogenous technological change”, Journal of Political Economy

94 (1), pp. 71-102.

Romer P. (1994), “The origins of endogenous growth”, Journal of Economic Perspec-

tives 8 (1), pp. 3-22.

Schmutzler A. (1999), “The New Economic Geography”, Journal of Economic Sur-

veys, 13 (4), pp. 355-379.

Scitovsky T. (1954), “Two concepts of external economies”, Journal of Political Econ-

omy 62, pp. 143–151.

Solow R. (1956), “A contribution to the theory of economic growth”, Quarterly Journal

of Economics 70 (1), pp. 65-94.

28
 Working Paper 16-02

Starrett D.(1978), “Market allocations of location choice in a model with free mobility”,
Journal of Economic Theory 17, pp. 21-37.

29