INDUSTRIAL AGGLOMERATION: ECONOMIC GEOGRAPHY, TECHNOLOGICAL SPILLOVER

AND POLICY INCENTIVES

Author(s): Emanuele Bracco
Source: Rivista Internazionale di Scienze Sociali, Anno 122, No. 1 (Gennaio-Marzo 2014), pp.
3-18
Published by: Vita e Pensiero – Pubblicazioni dell’Università Cattolica del Sacro Cuore
Stable URL: https://www.jstor.org/stable/43830198
Accessed: 17-02-2020 20:24 UTC

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/43830198?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Vita e Pensiero – Pubblicazioni dell’Università Cattolica del Sacro Cuore is collaborating

with JSTOR to digitize, preserve and extend access to Rivista Internazionale di Scienze
Sociali

This content downloaded from 45.238.196.189 on Mon, 17 Feb 2020 20:24:12 UTC
All use subject to https://about.jstor.org/terms
 Rivista Internazionale di Scienze Sociali, 2014, n. 1, pp. 3-18

INDUSTRIAL AGGLOMERATION: ECONOMIC GEOGRAPHY,

TECHNOLOGICAL SPILLOVER AND POLICY INCENTIVES

Emanuele Bracco*

ABSTRACT

In a Marshallian Industrial District (MID) agglomeration of firms is triggered by the pr

specialized workforce in a concentrated area. Labour mobility across firms then generat
edge spillover across firms, i.e. a positive location. This centripetal force is balanced by
gal forces like congestion costs and increased prices of specific factors caused by the inc
cal demand. On the other hand, the New Economic Geography (NEG) approach on firm
focuses on pecuniary advantages deriving from being clustered. This implies that a
bound to have specific welfare advantages when an industrial cluster is located within it
of these advantages, there may be incentives for "industry grabbing" policies and tax com
between nations or regions. We analyse the case of policy incentives in a NEG framew
technological spillover to the R&D sector. In this framework subsidizing R&D in the indu
gion may lead to welfare improvement.

Keywords : Industrial districts, Subsidy, New economic geography, Industrial cluster.

JEL Classification : 030, 038, RIO, R12.
u
I
i
C/3 1. INTRODUCTION

I Since Alfred Marshall's (1920) seminal work, the study of industrial clus
u
has been a central theme in Industrial Economics. Nevertheless only rece
:S attention has been brought back to this topic, thanks to a renewed interest in s
ing the origins and dynamics of geographical distribution of firms and
I
•S
brought by New Economic Geography (NEG).
The original Marshallian Industrial District (MID) is an agglomeration o
lar firms, often of small and medium size. The lack of strong economies
I and the presence of a specialized local workforce, which is highly mobile
firms of the same districts, generates knowledge spillovers and location ex
ities. This is often coupled with a high level of trust and personal interact
is more likely to be found in geographically small areas.
1
£
4>

¿3
>
TÍ" I would like to thank Professor Steven Redding for the guidance in writing this art
S Professor Maurizion Motolese for the incouragement and mentoring. The usual disclai
© plies.
* Emanuele Bracco, Department of Economics, Lancaster University, Lancaster, LAI 4YW,
United Kingdom. Email: e.bracco@lancaster.ac.uk.

This content downloaded from 45.238.196.189 on Mon, 17 Feb 2020 20:24:12 UTC
All use subject to https://about.jstor.org/terms
 4 E. BRACCO

The New Economic Geo

trial agglomeration assu
Monopolistic Competiti
locate close to other fir
lized labour force, i.e. l
of pecuniary advantage
duced by other firms, pr
er transport costs. The N
that are not necessarily
can be easily applied to
In such a setup, havin
terms of higher social w
policies may arise, whic
gain the presence of a d
This issue is particular
the European Union the
clusters. These "Smart
analyse what are the lo
funds for subsidizing res
Generalizing and expa
model in which the ma
technological spillover,
fare effect of local subs
In Basevi and Ottavian
externalities, the equilib
both from the District a
the possibility of subsi
ments, sub financing the
that low level of subsid
In section 2 contains t
model, Section 4 explor
tion 5 concludes.

2. REVIEW OF LITERATURE

This paper is related to the New Economic Geography literature of ind

agglomeration, which is largely based on a monopolistic competition mark
ture à la Dixit and Sitglitz (1977). According to this seminal paper, the ag
tion force, given by increasing returns to scale at the firm level, is bal
consumers love for variety modelled through a CES utility function and
graphic dispersion of demand.
One of the main contribution in this strand of literature is Krugman a
ables (1996), which is based on a two-country and two-industry model, wi

This content downloaded from 45.238.196.189 on Mon, 17 Feb 2020 20:24:12 UTC
All use subject to https://about.jstor.org/terms
 INDUSTRIAL AGGLOMERATION 5

industries modelled à la Dixit and Stiglitz (1977), in

migrate between the two locations. Migration movem
ers seeking to move to the locality with higher real
trial agglomeration occurs because of the presence of
use output of other firms as input, and in particular
duced by firms belonging to the same industry, whil
supplied by firms in the other sector. Because of th
any firm will have the advantage to stay as near as
is constituted not only by workers' demand for the fin
mand for intermediates from firms, with intra-industr
dustry trade.
Because of this features, firms tend to cluster together to save in transport costs.
This in turn leads to ail increased demand of labour, which triggers migration into
this region, increasing again the demand in the region. Agglomeration stops occur-
ring when the savings in trade costs originated through the intra-industry trade are
not enough to counterbalance the (trade) costs of exporting goods to the other
country. When trade costs are sufficiently high, the advantage of being near to si-
milar firms is overtaken by the costs of exporting.
The equilibrium outcome of such a model is that for higher values of trade
costs the unique equilibrium is a symmetric one in which both countries have both
industries and the real wages are equalized. As trade costs get smaller the outcome
changes dramatically, with the insurgence of two (stable) agglomeration equilibria
and one (less stable) symmetric equilibrium. Which of these equilibria is actually
realized will crucially depend on the starting situation. If the starting point is a si-
tuation sufficiently balanced between the two regions, this will lead to a symmetric
equilibrium. If instead the starting point was already unbalanced in some way, this
unbalance will be self-reinforced and will lead to a polarized agglomerated equili-
brium. The actual result crucially depends on parameter values, such as the
strength of inter-industry and intra-industry linkages. As trade costs diminish
further the clustering equilibria will be reinforced: there will still be a symmetric
equilibria, but it will become unstable, and any small perturbation will lead to a
clustering equilibrium in either of the two regions. The prevalence of agglomerat-
ing equilibria for low enough values of the trade costs hinges mainly on the as-
sumption that intra-industry linkages are stronger that inter-industry ones. Vice ver-
sa the opposite will be true in equilibrium, with firms having an incentive to locate
close to firms belonging to the other industry, and dispersion prevailing.
Venables (1999) analyses a scenario in which there are pecuniary advantages in
hosting or not a district, and consequently there are incentives for industry grab-
bing policies. This paper describes a multi-sector model, in which all industries
but one produce tradable goods à la Dixit and Stiglitz (1977), while a non tradable
good is produced by a perfectly competitive constant-returnto-scale sector. Produc-
tion inputs are labour and output of other industries. The strength of intermediate
linkages, trade costs and the nature of the increasing returns to scale determine
whether centripetal or centripetal forces are to prevail.

This content downloaded from 45.238.196.189 on Mon, 17 Feb 2020 20:24:12 UTC
All use subject to https://about.jstor.org/terms
 6 E. BRACCO

Starting from this for

we expected, of complete
When trade costs are sm
indeterminate. It is poss
the equilibrium allocatio
nacy will be larger for ce
industry) linkages becom
From the welfare poin
having a higher level o
Home, in addition to savi
eign to Home. That is w
number of industries and
level of trade costs, in
one Nation's welfare is a
The focus of this paper
dustries, which are fina
fare point of view these
they show diminishing r
An analysis of how a r
developed in Venables an
ing if a tax on one indust
in a particular location,
this relocation is perman
Starting from the usua
partial equilibrium appr
el includes well-known
duction of industrial go
turn-to-scale sector. In
same analyzed in Krug
strength of the linkages
The effects of a tax on
dent on trade cost and
outcome of taxation is t
country without tax wi
of an agglomerated equ
of trade costs. The level o
it could in fact make un
the more favourable locat
In presence of strong i
so big that a stable asy
range in which it occur
situation will be less an
costs.

Also Kind et al. (2000) concentrated their attention on the effect of tax compe-

This content downloaded from 45.238.196.189 on Mon, 17 Feb 2020 20:24:12 UTC
All use subject to https://about.jstor.org/terms
 INDUSTRIAL AGGLOMERATION 7

tition on agglomeration and welfare in a NEG model.

country two-sector model with intermediate linkag
considering the presence of a second factor of prod
based on mobile capital and immobile labour; capi
abroad and moves towards the best rents. Taxes are le
base, and entirely redistributed back to consumer. In
keeps the well know features of on agglomeration, di
libria depending on the strength of intermediate lin
costs. In their setup consumers are be willing to subs
in transport cost for as many varieties as possible; on
of strong linkages decreases the incentives for a fir
breaking the existing agglomeration, even if it must
a positive tax. In presence of strong linkages the co
extract the localisation rent. As intermediate linkage
optimal for the core region to actually subsidize the
costs.

Baldwin and Krugman (2004) start from a similar setup of Kind et al. (2000): a
two-country, two-factor of production, and two-sector (manufacturing and agricul-
tural) model, with a market structure à la Dixit and Stiglitz (1977). In their model
each unit of capital represents an entrepreneur who is necessary for the production
of a certain variety. This implies that the number of varieties is exogenous, inde-
pendently of firms location. In this way capital will gain an agglomeration rent
due to the increasing return/transport cost properties that can lead to a concentrated
equilibrium. They also assume a government that does not just maximize citizens
welfare, but receives positive utility from tax revenue - that are considered a proxy
for political rents - and disutility from the distortionary effect of taxation, taking a
more disenchanted view on the role of government. The tax-competition equili-
brium strategies are such that the core extract a location rent from firms, and sets a
tax level which is low enough to avoid firms to be pushed to move to the periph-
ery. At the same time periphery wants to offer incentives to recall industries to his
location. The equilibrium outcome is a Stackelberg-like equilibrium in which the
core levies a higher tax on its citizen, but low enough that periphery does not have
the incentive in attempting an industry-grabbing policy.
This finding also tells us that a tax harmonization will be detrimental for both
countries: to impose a median level of tax will not allow periphery to recall firms,
and will set a too high and welfare diminishing tax rate for periphery itself. At the
same time the core will have a too low tax rate that, if increased, would increase
also government utility. A tax cap would have the same effect; a tax floor could
instead be weakly Pareto improving: it would not allow periphery to steal industry
and improve the core's welfare.
Baldwin and Okubo (2006) explored the effects of considering firm-level het-
erogeneity in a NEG model. The starting point is a so-called footloose capital
model. As in Baldwin and Krugman (2004) every firm needs a unit of capital, that
is inter-regionally mobile (the "entrepreneur"). The number of varieties is there-

This content downloaded from 45.238.196.189 on Mon, 17 Feb 2020 20:24:12 UTC
All use subject to https://about.jstor.org/terms
 8 E. BRACCO

fore exogenously set. Het

tivities of firms. Each uni
distributed according a P
any market, so that each
and there is a relocation cost in case one decides to relocate ones firm in the other
country. As one would expect, capital moves towards better nominal rewards. The
presence of the well-known agricultural CRS perfectly competitive sector equalizes
wages across countries.
As relocation costs decrease, the first firm to move towards the nation with a
larger manufacturing sector (and thus higher profitability) is the more productive
firm, i.e. the one that can gain more from moving. As a consequence, a cut-off le-
vel of inefficiency will be determined. At intermediate trade costs, as the first to
move are more productive, the agglomeration process is usually more limited than
without heterogeneity, as less relocating firms will be needed to reach the equili-
brium. At very high or very low trade costs heterogeneity doesn't affect in any
way agglomeration and dispersion forces. In this situation a hypothetical subsidy
introduced by one economy will have peculiar consequences. As we expect higher
subsidy will attract a higher number of firms, and subsidies will be more effective
as trade gets freer. The final outcome of this is a sorting equilibrium in which best
firms will remain in the district, while the less productive will migrate to the rest
of the world.
This paper also contributes to the new growth theory, as long as we think of
district as any endogenously growing locale. In this sense our contribution can be
linked to Romer (1990) and Grossman and Helpman (1991), as we describe the re-
search labs as entities which take advantage of location externalities.
The closest paper in this strand is Souberayn and Thisse (1999), who modelled
a partial equilibrium framework with learning by doing 'a la Arrow (1962) with
workers regionally immobile. In their model, the industry has increasing returns to
scale at the district level, which depend on the experience accumulated by workers
at that locale; in this way the technological spillover is formalized, so to give a
production advantage to localities where more output has been produced in the
past. This implies that firms face decreasing returns in the short run but increasing
returns in the long run. This dynamic model is constructed in a way such that en-
trepreneurs at the beginning of each period must decide where to locate, and then
hire capital in the world market and labour in the local market. The agglomeration
forces at work, as we already said, are dependent on the increasing returns at the
district level, while the dispersion force is due to the immobility of labour: a larger
presence of firms in a locale will determine a harder competition for workers, and
consequently higher wages. Locale with larger knowledge will have a larger num-
ber of firms, as well as smaller labour input requirements and higher wages. The
focus of the cited paper is on the dynamic evolution of districts, i.e. the long run
equilibrium in a world with two districts differently endowed with knowledge. The
equilibrium outcome may either be one in which history prevails, with the locale
endowed with more intial knowledge prevails, or one in which the two district

This content downloaded from 45.238.196.189 on Mon, 17 Feb 2020 20:24:12 UTC
All use subject to https://about.jstor.org/terms
 INDUSTRIAL AGGLOMERATION 9

converge in the long run. The final outcome is that the

how has more firms, more profits and higher wages, so
will be "richer". This also means that the number of fir
and profits) depends not only on ones own level of know
vel of knowledge of every other district. Furthermore,
or in the international cost of capital can easily imply
new districts. This is a theoretical finding that perfectly
namics of this particular form of industrial organization.

3. THE MODEL: NEG-TYPE MODEL

WITH TECHNOLOGICAL SPILLOVER

Our analysis continues on the line traced by Basevi and Ottaviano (2

which analizēs industrial clustering trying to bring together the modeling
works of the new economic geography and of the Marshallian district liter
Our aim is to study the welfare effects of a subsidy to the R&D sector.
The economy consists of two regions; in one of them a district is already
sent. The district is a cluster of horizontally differentiated firms à la Dixit a
glitz (1977). Each firm produces potentially for both the home and the fo
markets, and has to bear trade costs to sell to the distant market. Trade cos
modelled as iceberg costs. Labour L is distributed within locations such th
time t L(t) = XL workers are in the district. Workers are immobile betwee
gions and are employed in three sectors: a horizontally differentiated manuf
ing sector D , which produces N(t) different varieties out of which n(t) in t
trict, a costant return to scale sector Y, which produces a homogenous
whose price will be taken as numeraire and whose labour input requiremen
unit as well, and a R&D sector. The latter produces knowledge capital, emb
in blueprints protected by infinitely lived patents. Every firm needs to "
blueprint to produce a certain variety. The innovation sector is characterize
positive externality, by which productivity is an increasing function in the
already produced. Consumers maximise their intertemporal utility function
/*00

u= In 'D{t)aY{t)x-a)e->" dt (1)
Jo

[ fNW
with D{t) - / c(s, t) ° (2)
Uo J

where c(s, t) is the consumption of a

p > 0 is the rate of time preference
substitution between varieties and elast
Y is necessarily produced in both re
sector consists of perfectly competit
spillover. Productivity is an increasin

This content downloaded from 45.238.196.189 on Mon, 17 Feb 2020 20:24:12 UTC
All use subject to https://about.jstor.org/terms
 10 E.BRACCO

"present" in both regions

sion of labour input coef
knowledge variable.
The focus of this paper is
why it is made a simplify
the district and, because
in the district. Moreover i
tion of a blueprint abroad
this is to underline the lo
capital embodies a bluepr
l/</> unit of capital, wit
knowledge embodied in blu
These characteristics of
ditional theories of Industr
ture defines districts as
curs through interchange
mobile) specialized workf
has to do with technologi
the one analyzed in our
technology, but more of
tellectually" stimulating e
similar products.
The number of varieties
and N(t) = [7 (t) + </>[l -
Rest of the World.
The equilibrium allocation of resources is driven by financial markets where a
risk-free bond with interest rate r(t) is traded and from which investment in R&D
are financed.
Agents maximize their utility subject to the following budget constraint:

À(t) = r(t)A{t ) + w(t) - E(t) (3)

where A(t) is the wealth detained in safe bond, w(t) is the wage level, an
individual expenditure

fN(,)
E(t) = p{s,t)c(s,t)ds+pYY(t ) (4)
Jo

where p(s, t) is the price of variety s at time t, and c(s, t ) is the quantity consumed
of variety s at time t. As expected, the equilibrium result is the usual result from a
Euler equation in Ramsey models:

E(t) E*(t )

Th

This content downloaded from 45.238.196.189 on Mon, 17 Feb 2020 20:24:12 UTC
All use subject to https://about.jstor.org/terms
 INDUSTRIAL AGGLOMERATION 1 1

impatience. Demand for home and foreign produced varieti

c(i, t ) = Pp^ý-a a E(t)XL , i e [

c(j, t) = q^_l aE(t)XL, j e [i

where P(t) indicates the usual NEG price index:

" fM{t) ì-rb" r /•«(') rm "ļ-rb"

p(t)= .JO
/ P(s>()J°dsUo
= / P(M)
Jn(t)
a)+ p{j,t) adj (8)
Results for the Rest of the W
tion and free trade in the agric
and leads them towards unity, th
The firms' objective function c

II = pc + q* c* - ß(c + re*) - R (9)

II* =p* è* +q* c* -ß{c + rc*)-1

where is the rental rate of the blueprint.
We can derive the classic Dixit and Stiglitz (1977) results in w
fixed and all firms generate the same profits.

P=r-=V^Ī <">
* í i 'i '

Q = Q * =rļ~r í (12) i 'i '

*-£r <l3)
In equilibrium the profits are going to be zero, and
in (13) must be equalized to the rental rate of capita
the value of a unit of knowledge capital embodied in
the present discounted value of rental rates:
poo
v(/)= / tf(5)ír[r(í)-''(')l ds (14)
JO

and innovation activity will be regulated by an arbitrage equation between the in-
vestment in a safe bond and investment in innovation

tv =
v
- = - (15)

This content downloaded from 45.238.196.189 on Mon, 17 Feb 2020 20:24:12 UTC
All use subject to https://about.jstor.org/terms
 12 E.BRACCO

Following the custom in th

on the case in which expe
grows at a constant rate

_k-Ñ-¿
8 _ K - N - V ( }
The final condition that needs to be imposed is the mar
i.e. equalizing labour supply and demand in each sector, tog
equation (15) calculated in steady state.

S=~[A+(l-A)r]¿-/> (17)

Z = ļf+£T^[A + (1
Thus they derive an equilibrium value for world e
rate solving (14) and (15) for ['E + (1 - A)£*]

[XE + (l-X)E*}
¿7
= l+-Ç- (19)
aL 7 cr - a . x
áT = - -r
cr r] a

The result of equation (20) tells us that there w

too small, i.e. if 7 is smaller than a certain th
how a district must have some kind of critica
tion and therefore to grow.
Total expenditure can be split between the MID
diture, given that by assumption no one in the r

'E = 'SL (21)

Lj

(I - ')E* = (I - X) (22)

In particular in equation (21) we can distingu

bill, and a second term that represents the w
prints inherited from history. With all these
state value of 7, i.e. the dimension of the distric
It is defined an indirect utility function as t
at time zero plus the utility deriving from the i

(23)

In presence of technological spillover, as we expect, market allocates resources

inefficiently: in particular it allocates too little resources to the innovation sector

This content downloaded from 45.238.196.189 on Mon, 17 Feb 2020 20:24:12 UTC
All use subject to https://about.jstor.org/terms
 INDUSTRIAL AGGLOMERATION 1 3

and too much to production. The derivative of the indir

respect to 7 is always increasing in the interval [71, 1]. M
ment turns out to be always welfare decreasing. Specific

dV a2L 1 /? | a 1 - 6<ļ>
dy fP-rja(a - 1 ) 7 ^XL-^prj p(a - I) 7 + <5</>(l - 7)
Where the first term on the RHS captures the positive externality of 7 on the
growth rate, the second term captures the negative effect of growth on blueprint
value and thus on wealth, the third term is a positive component due to the savings
in trade costs as the share of varieties produced in the district increases.
Given this outcome we can now turn into analyzing what are the effect on the
equilibrium and the welfare implications of introducing a subsidy to the R&D sec-
tor, financed through a lump sum tax on all workers in the district.

4. POLICY AND WELFARE IMPLICATIONS OF A SUBSIDY TO THE R&D SECTOR

The main contribution of this paper is the analysis of the welfare e

subsidy ip G [0, 1] on the wage of every worker in the Innovation sect
through a lump sum tax on each agent in the district. In other words
know the effect of R&D support policies in a two-region setting in c
vancement in knowledge "leak" to the neighbouring firms, generating
ternalities.
Due to this subsidy, we expect the growth rate to increase, more re
be devoted to the Innovation sector, and the equilibrium to be closer to
allocation outlined in the previous section.
The value of a single innovation laboratory, that without subsidy is v
becomes equal to = 77(1 - ^)/(t K): an increase in the number of bl
the economy brings a decrease in the present discounted value of their
In fact we can confirm that the growth rate of knowledge remains equ
gative growth rate of R&D firms:
v

8 = ~u
From the arbitrage equation (15) we can derive a new expression for the steady
state growth rate:

8» = 7I«^)|A£+t'-x,£-1 <25>
This together with the unchanged labour-market clearing condition (18) gives
the result of a new steady state growth rate and expenditure result:

ia£+i 'i - ^ (.-«o -7) +<*) (26)

This content downloaded from 45.238.196.189 on Mon, 17 Feb 2020 20:24:12 UTC
All use subject to https://about.jstor.org/terms
 14 E.BRACCO

aí 7 1 -V>)(q-g)
^ cr( 1 - -0) + atjj r' cr(l - ip) + aip
From these results we can note how equation (27) shows an in
growth rate. The presence of a free trade sector always present
allows us to have an equilibrium wage which remains equal to un
tries. Moreover, if in the Rest of the World no innovation activi
in absence of the subsidy, we can fortiori state that this activity wi
in presence of a subsidy to R&D in the district. From this we c
the result stated in (22) (i.e that that (1 - ')E* = (1 - A)) still ho
Expenditure by MID residents therefore is:

This amount is decreased with respect to the case in which the

as in (21). This is due both to the cost of the subsidy born by th
dents and to the negative effect of the growth rate on the valu
blueprints inherited by history.
We define the total tax bill TB as the product of labour input requ
novation sector, R&D production (i.e. g ) and the subsidy i
rewrite the total expenditure in the district by residents A E^L as:

A E^L = XEL - TB - -tpp (29)

7

From this formula one can notice how expenditure is affected both by the tax
bill and by the decreased value of the stock blueprints inherited by history.
The minimum size of the district, which is necessary for growth to take place,
changes as well. There is growth, stemming from the subsidized innovation sector,
if the share of capital in the district is at least

that is clearly negatively related to the size of the subsidy ip. This means that a po-
licy that encourages research through subsidies to the Innovation sector decreases
the chance for a cluster to die out as a centre for innovation for its lack of critical
mass.

As for the welfare effects of the subsidy, the implicit expression o

(23) remains unchanged:

but its value has changed, due to the effects of the subsidy on ex
growth rate ip.

This content downloaded from 45.238.196.189 on Mon, 17 Feb 2020 20:24:12 UTC
All use subject to https://about.jstor.org/terms
 INDUSTRIAL AGGLOMERATION 15

Let us now analyze how the welfare changes at the chang

dV a2L
dy (P-rp[a{' - ip)

The signs of three components that would be there also in absence of a subsidy
(see equation [24]) are unchanged. Nevertheless both the effect through growth
(first term) and the one through expenditure (second term) are changed in size.
The growth effect is clearly increased, to state the "subsidised" externality at
work. Nevertheless we can just state that at least in case of low subsidies this va-
lues is still increasing in the relevant range of 7.
To calculate the optimal subsidy, we must analyse the partial derivative of wel-
fare with respect to the subsidy itself.

dV _
dip

¿¿(f-l)

It can be seen how in the positive neighbourhood of ip = 0, i.e. in

small subsidy, the welfare effect is likely to be positive. It must b
again how the transmission mechanism relies on the acceleration of
process (first term on RHS) as a positive force, but is limited by th
fect on expenditure (second term on RHS).

5. CONCLUSIONS

Previous literature has showed that policy interventions when the in

structure is not perfectly competitive have positive effects on welfare a
give start to a race to the bottom. Moreover tax harmonisation or tax ca
sily detrimental and not welfare improving. In NEG models, the transm
chanism relies mostly on subsidies allowing for transport cost saving or
propriating location rents. Through fiscal policy in fact citizens can take
of the presence of industrial agglomeration in their territories, either s
their presence or trying to extracting a location rent. In the more tradit
shallian models, policies instead correct the inefficiencies arising from ex
and thus allow the economy to reach the efficient production point. Th
analyzes the effects of tax policy in a model describing a NEG-type dist
technological spillover, and aims at shedding light on the efficiency and
of local subsidies to R&D-intensive firms, which are becoming increasing
lar in the EU. Our findings are that subsidies to the innovation sector ar
in increasing growth rates of innovation and in enlarging the possibiliti
district of maintaining its innovation activity (decrease in jL) in order t
Nevertheless resources moved from taxpayers to innovation firms and

This content downloaded from 45.238.196.189 on Mon, 17 Feb 2020 20:24:12 UTC
All use subject to https://about.jstor.org/terms
16 E.BRACCO

creased value of ideas ha

highlighting important
we can state that for sm
subsidising the innovatio

This content downloaded from 45.238.196.189 on Mon, 17 Feb 2020 20:24:12 UTC
All use subject to https://about.jstor.org/terms
 INDUSTRIAL AGGLOMERATION 1 7

REFERENCES

Arrow K.J. (1962). The economic implications of learning

Studies , 29, 152-173

Baldwin R., Krugman P. (2004). Agglomeration, integration

mic Review , 48(1), 1-23

Baldwin R., Okubo T. (2006). Heterogeneous Firms, Agglom

phy: Spatial Selection and Sorting. Journal of Economic Ge

Basevi G., Ottaviano G. (2002). The district in the global e

reign location. Journal of Regional Science , 42(1), 107-126

Dixit A., Stiglitz J. (1977). Monopolistic Competition and O

American Economic Review , 67, 297-308

Fujita M., Krugman P., Venables A. (1999). Spatial Economy

national Trade. MIT Press, Cambridge MA, USA

Grossman G., Helpmann E. (1991). Innovation and Growth

Press, Cambridge MA, USA

Jones C. (2002). Introduction to Economic Growth , W.W. N

Kind H.J, Midelfart-Knarvik K.H., Schjelderup G. (2000). Co

py world. Journal of Public Economics , 78, 253-274

Krugman P., Venables A. (1996). Integration, specializat

Economic Review , 40, 959-967

Krugman P., Venables A. (1995). Globalization and the ineq

Economics , 110, 857-880

Krugman P. (1991). Geography and Trade. MIT Press, Camb

Marshall A. (1920). Principles of Economics, 8th edition. M

Norman V.D., Venables A. (2004). Industrial Clusters: Equil

Economica , 71, 284, 543-558

Ottaviano G., Puga D. (1998). Agglomeration in the glob

"new economic geography". The World Economy, 21(6), 70

Romer D. (1990). Endogenous Technological Change. Journ

S71-S102

Souberayn A., Thisse J.-F. (1999). Learning-by-doing and the Development of Industrial
Districts. Journal of Urban Economics , 45, 156-176

This content downloaded from 45.238.196.189 on Mon, 17 Feb 2020 20:24:12 UTC
All use subject to https://about.jstor.org/terms
18 E.BRACCO

Venables A. (1996). Equilib

Economic Review , 37, 341-

Venables A. (1999). The inte

vantages in a multi-industr

Venables A., Neary P.J. (1

in location. In R. Baldwin
Cambridge University Pres

This content downloaded from 45.238.196.189 on Mon, 17 Feb 2020 20:24:12 UTC
All use subject to https://about.jstor.org/terms