This article was downloaded by: [Nipissing University]On: 06 October 2014, At: 05:46Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number:1072954 Registered office: Mortimer House, 37-41 Mortimer Street,London W1T 3JH, UK

International Journal ofthe Economics of BusinessPublication details, including instructions forauthors and subscription information:http://www.tandfonline.com/loi/cijb20

MNCs, TechnologicalInnovation and RegionalSystems in the EU: SomeEvidence in the Italian CaseJohn Cantwell & Simona IammarinoPublished online: 21 Jul 2010.

To cite this article: John Cantwell & Simona Iammarino (1998) MNCs,Technological Innovation and Regional Systems in the EU: Some Evidence in theItalian Case, International Journal of the Economics of Business, 5:3, 383-408,DOI: 10.1080/13571519884459

To link to this article: http://dx.doi.org/10.1080/13571519884459

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of allthe information (the “Content”) contained in the publications on ourplatform. However, Taylor & Francis, our agents, and our licensorsmake no representations or warranties whatsoever as to the accuracy,completeness, or suitability for any purpose of the Content. Anyopinions and views expressed in this publication are the opinions andviews of the authors, and are not the views of or endorsed by Taylor& Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information.Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilitieswhatsoever or howsoever caused arising directly or indirectly inconnection with, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private studypurposes. Any substantial or systematic reproduction, redistribution,reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of accessand use can be found at http://www.tandfonline.com/page/terms-and-conditions

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

International Journal of the

Economics of Business, Vol. 5, No. 3, 1998, pp. 383± 408

MNCs, Technological Innovation and Regional

Systems in the EU: Some Evidence in the Italian

Case

JOHN CANTWELL and SIMONA IAMMARINO

ABSTRACT The increasing appreciation of the role of multinational corporations

(M NCs) in the generation of technology across national boundaries has been facilitated

by the recent trend for MNCs to establish internal and external networks for innovation.

The development of cross-border corporate integration and intra-border inter-company

sectoral integration makes it increasingly important to examine where and how innovative

activity by MNCs is internationally dispersed and regionally concentrated. By using

patents granted to the largest industrial ® rms for innovation located abroadÐ arranged by

the host region Ð we test the nature of the relationship between the foreign-owned and the

indigenous company pro ® les of technological specialisation in the Italian regions. We

argue the MNC networks for innovation in Europe conform to a geographical hierarchy

of regional centres. Accordingly, the technological specialisation of foreign-owned af® liates

in different regional locations depends upon the position of the region in the hierarchy, i.e.

whether the regional system is at the top of the hierarchy (higher order location) or is a

lower order regional centre.

Key words: Multinational corporations; Technological innovation; Regional sys-

tems.

JEL classi® cations: O1, O3, R1, F2.

Introduction

The evolutionary concept of technological innovation as a `knowledge-accumulat-

ing’ and `location-speci® c’ process, along with the intense debate on the globaliza-

The authors are grateful to participants at the International Conference on ª Technology Policy and Less

Developed Research and Development Systems in Europeº held in Seville in October 1997 for helpful

comments on an earlier draft. Simona Iammarino wishes to acknowledge the support of the EU

Commission under the TMR Marie Curie Research Training Programme (Contract

No. ERBFMBICT961062) .

John Cantwell and Simona Iammarino, University of Reading, Department of Economics, Whiteknights, PO Box

218, Reading RG6 6AA, UK. Fax: 1 44 (0) 1189 750236 ; e-mail, k J.A.Cantwell@reading.ac.uk l ,k S.Iammarino@reading.ac.uk l .

1357-151 6/98/030383-2 6 $7.00 Ó 1998 Carfax Publishing Ltd

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

384 J. Cantwell and S. Iammarino

tion of innovation, have recently drawn attention to multinational corporations

(MNCs) as creators of innovation across national boundaries.

The increasing appreciation of the role of MNCs in the generation of technol-

ogy (and not just its international transfer) has been facilitated by the recent trend

for MNCs to establish internal and external networks for innovation. Internation-

ally integrated networks within the ® rm may lead, through a greater focus in the

specialisation of technological efforts in af® liates, to an improvement of innovation

capacity both of the MNC and of the host location. Inter-® rm networks estab-

lished between MNC af® liates and local ® rms may, in addition, amplify the

advantages of geographical agglomeration in some particular lines of technological

development.

In other words, the development of cross-border corporate integration and

intra-border inter-company sectoral integration makes it increasingly important to

examine where and how innovative activity by MNCs is internationally dispersed

and regionally concentrated. This is particularly relevant in the case of the

European Union (EU), where the internationalisation of technological activities

within the area has been spurred even more intensively by the adjustment made by

MNCs in response to the deepening of the economic integration process over the

1990s.

In this paper we argue that MNC networks for innovation in Europe conform

to a geographical hierarchy of regional centres, the precise composition of which

differs between industries. Accordingly, the technological specialisation of foreign-

owned af® liates in different regional locations depends upon the position of the

region in the hierarchy, i.e. whether the region is at the head of the hierarchy

(higher order location) or is a lower order regional centre.

The paper is divided into ® ve parts. The relevance of `the local vs the global’

in innovative processes is the subject of the ® rst section. In it we explore the

change of MNC innovatory strategies and the `cumulative causation’ mechanism

which may operate in different regional locations, according to the technological

features both of the MNC and of the regional system. The search for a more

appropriate geographical indicator and the methodology for the regionalisation of

the patenting activity for innovation undertaken abroad by the world’ s largest

industrial ® rms are discussed in the second section. By using patents granted to the

largest ® rms in the US and classi® ed by the location of the research facility (within

the ® rm) responsible for the inventionÐ arranged here by the host European

regionÐ section 3 explores the general features of the location of foreign research

in the Italian case. Section 4 provides a preliminary attempt to test the relationship

between foreign-owned and indigenous ® rms’ pro® les of technological specialis-

ation in different Italian regional systems. Finally, the ® fth section presents some

concluding remarks and highlights the direction of our future research.

1. Local and Global

1.1. Why do Regions Matter?

The increasing interest of economists in the globalisation1 of production and

technology has reached its peak with the approaching of the new millennium.

However, the dif® culty of assessing the actual `state of the art’ and the implications

of globalisation, especially as far as innovative phenomena are concerned, has led

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

M NCs, Innovation and Regional Systems 385

to rather dissimilar views, still inadequately supported Ð at least in most casesÐ by

signi® cant empirical evidence.2

The expression ª globalisation of innovationº is used precisely to describe the

recent increase in the intra-® rm coordination of the innovative activities of MNCs.

On the one hand, the strength that allows a ® rm to invest and govern its operations

across national boundaries is its ability to innovate and to take advantage of such

innovation in different locations through its own organisation. The authentic

global generation of innovations requires, on the other hand, a wide range of skills

and capabilities that only ® rms with speci® c infrastructure, organisation and

management can attain (Archibugi and Iammarino, 1998). This explains why

there has been a shift in attention away from the MNC as a mere vehicle of

technology transfer towards the crucial role it plays as a creator of innovation and

technological knowledge (Chesnais, 1988; Cantwell, 1994).

The `state of the art’ on the international generation of innovationÐ leaving

aside the current academic debate3- may be depicted as a trend towards increasing

shares of innovation generated outside the home country and integrated within the

MNC. The share of US patents of the largest industrial ® rms attributable to

research undertaken in foreign locations rose from 12.2% in the period 1969± 72

to 18.6% in the years 1987± 90 (Cantwell, 1995, tables 1± 2). In the latter period,

moreover, European parent ® rms showed on average a share of patents granted for

research located outside their home country of 30.9%, indicating a greater

propensity to internationalise their innovatory capacity than those of US origin

(whose share over the same period was 8.9%) or Japanese (slightly above 1%).

The traditional advantages of the centralisation of research and innovative

activitiesÐ basically connected to economies of scale and scope in R&D, control

over innovation, and linkages with national business and non-business sectors Ð

seem thus to be increasingly counterbalanced by those associated with

decentralisation (see Pearce and Singh, 1992; Howells and Wood, 1993; Miller,

1994). From the perspective of the investor, the latter can be summarised in terms

of the linkages between innovatory activity and foreign production, local markets,

suppliers and clients, and the exploitation of technological ® elds of excellence in

host countries.

Turning to the issue of the implications of the globalisation process, the

international dispersion of the creation of new technology and the change of

innovatory strategies of MNCs make it all the more important to take into account

the geographical concentration of MNC technological operations at a subnational

level.

It makes sense to assume that globalisation implies that location matters even

more than in the past. The signi ® cance of the `regional dimension’ of an innovative

system has emerged as the logical consequence of the interactive model (Kline and

Rosenberg, 1986), which put the emphasis on the relations with information

sources external to the ® rm. Such relationsÐ between ® rms and science infrastruc-

ture, between producers and users at inter-® rm level, between ® rms and the

institutional environmentÐ are strongly in¯ uenced by spatial proximity mecha-

nisms that favour processes of polarisation and cumulativeness DeBresson, 1987;

Lundvall, 1988; von Hippel, 1989). Furthermore, the employment of informal

channels for knowledge diffusion (the so-called tacit or uncodi® ed knowledge)

provides another argument for the tendency of innovation to be geographically

con® ned (HaÈ gerstrand, 1967; Lundvall, 1992). Although a break has thus oc-

curred with the conventional economic approach4 Ð in which spatial factors shap-

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

386 J. Cantwell and S. Iammarino

ing innovation were usually considered secondary (if not thoroughly negligible)Ð

the regional scope is still rather indeterminate with respect to the geographical

dispersion of innovatory capacity in the global economy. For this purpose, a

distinction is to be made between forces that tend to make related ® rms and

industries cluster spatially and forces that might lead to patterns of uneven regional

development, i.e. the emergence of centres and peripheries at the global and

national level (Malmberg, Solvell and Zander, 1996).5 Yet, the major risk of

regarding the innovative process as non-speci® c to the locational context in which

it is developed is to underestimate, or even to overlook completely, the distribution

of costs and bene® ts triggered by the globalisation of innovation.

This is all the more true in the presence of an in-depth process of economic

integration, as is the case of the EU, which creates the need to de® ne the problems,

and the policies aimed at solving them, in terms of geographic location and

centre/periphery economic convergence. The problem of the regional imbalances

becomes how to keep pace with the structural changes in the generation of

innovation due to both integration and globalisation processes. In fact, the

importance of the spatial dimension of innovation, in the new context, becomes

something of a paradox (Iammarino, Prisco and Silvani, 1995). At ® rst glance, it

may appear that by removing the barriers between European markets, encouraging

the spread of the new information technologies, improving telecommunication

systems, etc., the direction of change is one way, i.e. it has helped lessen

geographical constraints and localisation. On the other hand, as the creation of

new networks of transactions is extremely sensitive to geography, the knowledge

base matters increasingly at local/regional level, and economic growth requires the

creation and improvement of endogenous capabilities, which are highly path-de-

pendent and geographically-speci® c. Therefore, the centrality of local forward and

backward linkages and the emergence of a `performance’ type of MNC, identi® ed

by `heterarchical’ internal structure and innovation-based competitiveness (Dun-

ning, 1993; Cantwell, 1992a, 1994; Amin and Tomaney, 1995), tend to make the

geographical polarisation of innovative activities stable and self-organising. Insofar

as innovation is essential to any explanation of growth differential, it is clear that

a central role is played by regions as basic units in any cost/bene® t evaluation of

EU economic integration and in explaining the slow-down of economic conver-

gence within the area in the last decade.6

Hence, as both globalisation and economic integration processes are likely to

interact with the change of the MNC’ s organisation of innovative activitiesÐ spur-

ring the rising function of local innovative contexts Ð we could argue that MNCs

are to be considered as the key-ring of the chain from global to local.

1.2. MNC Networks for Innovation and Regional Location

The change in MNC strategies towards a greater degree of cross-border coordi-

nation of their internationally dispersed operations requires an organisational

structure that could not have been developed through purely arm’ s-length market-

based coordination between geographically separated units. As Dunning and

Robson (1987) have effectively shown, transaction costs may shrink with inte-

grated governance of units, whatever units are considered to be, either af® liates or

different locations. However, the transaction costs approach, focusing basically

upon bene® ts in terms of short-term ef® ciency and ¯ exibility, fails to take into

account the knowledge accumulation effects linked to the integration of innovative

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

M NCs, Innovation and Regional Systems 387

activities across geographically dispersed units. Dunning and Wymbs (1997) have

demonstrated that the degree of multinationality is signi® cantly associated with the

perception that ® rms increase their global technological advantage from foreign

sources. They pursue this aim by establishing geographically dispersed networks of

af® liates, as a means of building a sustainable competitive advantage based much

more on capabilities and dynamic improvements than on static ef® ciency factors.

Through such networks, technology, skills, and assets are transferred across

national borders in a two-way directionÐ from the parent to subsidiaries and from

af® liates to the parent company.

Such networks, however, take a rather complex con® guration precisely because

of the different types of agglomeration forces which shape spatial organisation. On

the one hand, there are general external economies and spillover effects which

attract all kinds of economic activities in certain regions and locations and

determine, in the case of corporate integration, the localisation of the new units.

These centripetal forces strengthen the inter-border intra-® rm integration and the

feedback of knowledge, expertise and information which occurs within networks of

af® liates. From the economic geography perspective, this leads to regional concen-

tration and thus to the emergence of regional cores with sectoral speci® cities

varying across different locations. On the other hand, localisation economies are

fostered in spatial clusters of related ® rms, where learning dynamics and exchanges

of tacit knowledge are embedded in a distinct environment of interactions among

different subjects. This kind of force is likely to be sector-speci® c and to produce

stable mechanisms of knowledge accumulation which enable host locations to

increase their competence over time. Thus, intra-border sectoral integration

implies local external networks between af® liates and indigenous ® rms and non-

market institutions. By tapping into local knowledge and expertise, foreign

af® liates gain a competitive advantage which can be exploited locally and/or

transferred back to the parent company, enhancing its global technological com-

petence. However, a geographical hierarchy of regional centres can be established,

as a consequence of the interaction and the intensity of general external economies

and localisation economies, which in turn depend upon the characteristics of the

regional system considered.

It has been argued that the possible effects on local economies are both direct

and indirect (Dunning, 1994). As a general rule, the amount of innovation

generated ex novo by foreign af® liates of MNCsÐ which includes also the innova-

tive effects of their demand and cost linkages with indigenous suppliers and

customers and their impact on local market structure Ð minus the amount of

`diverted’ innovation (i.e. that which would have been generated in the absence of

MNCs), gives the net technology creation effect. Therefore, MNCs’ global gener-

ation of innovation may enhance the local innovative capacity, as much as, in the

wrong circumstances, it may weaken it (Archibugi and Iammarino, 1998).

By mutating and adapting the classi ® cation proposed by Ghoshal and Barlett

(1990), it is possible to summarise the different patterns of knowledge ¯ ows

backwards and forwards within and outside the MNC (Figure 1).

Internal (intra-® rm) networks are established either in the case of a high

centralisation of innovative activities in the parent company, which transfers

technology, experience and knowledge to af® liates following an innovation pattern

of `Centre-for-global’ ; or in the case of highly decentralised innovation undertaken

by af® liates in different locations, which are subsequently transferred to the parent

company and possibly applied to other units within the MNC, following a strategy

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

388 J. Cantwell and S. Iammarino

Ty

pe

Ge

og

ra

ph

ica

lIn

no

va

tio

n¯

ow

sIn

no

va

tio

nstr

ate

gy

Imp

ac

t

dir

ec

tio

na

mo

ng

un

its

(Gh

osh

al

an

dB

arl

ett

,1

99

0)

INT

RA

-FIR

MC

RO

SS

-BO

RD

ER

Fro

mp

are

nt

toaf®

liate

sC

en

tre-

for-

glo

bal

Bo

tho

nM

NC

an

d

(Co

rpo

rate

inte

gra

tio

n)

Fro

maf®

liate

sto

pare

nt

Lo

cal

-fo

r-g

lob

al

ho

stlo

cati

on

INT

ER

-FIR

MIN

TR

A-B

OR

DE

RB

etw

een

af®

liate

san

dL

ocal

-fo

r-lo

cal

On

ho

stlo

cati

on

(Sect

ora

lin

teg

rati

on

)in

dig

eno

us

®rm

sL

ocal-

for-

glo

bal

CR

OS

S-B

OR

DE

RB

etw

een

pare

nt

an

dG

lob

al-

for-

glo

bal

On

MN

C

oth

er®

rms

Fig

ure

1.

MN

CN

etw

ork

sfo

rIn

no

vat

ion

.

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

M NCs, Innovation and Regional Systems 389

of `local-for-global’ . While in the ® rst case the impact on host location may be

adverse due to the competitive erosion of the domestic market base of indigenous

® rmsÐ which leads to a reduction in the research of local ® rmsÐ in the latter case

(Local-for-global), the creation of innovation by af® liates is more likely to posi-

tively affect, through knowledge spill-overs, the general innovation capacity of the

region in which they are located.

External (inter-® rm) networks provide intra-sectoral integration between for-

eign subsidiaries and local ® rms. In this case, the local character of innovation in

the MNC’ s af® liate, which follows a Local-for-local pattern trying to tap into

indigenous capabilities, might add substantially to the strength of the host region

in which it is active. The technological specialisation of the host location is thus

likely to be reinforced, as well as the technological competence of the MNC, which

absorbs sizeable corporate assets through its internal corporate network and the

complementary Local-for-global strategy. The choice of location by the MNC

depends upon the number of regional centres and their relative position in the

geographical hierarchy, and upon the extent to which the MNC has developed a

strategy for technological diversi® cation through tapping into speci® c competences

in a range of different higher order and lower order locations (Cantwell, 1994).

Finally, external networks can be established by the MNC parent through

strategic alliances and collaborations with other MNCs in selected areas of activity.

The importance of such relationships for mutual exchanges of knowledge and

information is increasing rapidly (for a review of the indicators, see Archibugi and

Iammarino, 1997), but since they do not depend upon geographically-speci® c

characteristics, they will not be considered in the following analysis.

To conclude, inter-border corporate integration and intra-border sectoral

integration seem to strengthen technological linkages and specialisation between

regions, especially where ® rms implement a rationalising strategy in order to adapt

to a changing economic environment such as the European Union during the

1990s (Cantwell and Dunning, 1991).

1.3. Global Generation of Innovation and Cumulative Causation

A crucial element in the model of local accumulation of knowledge involves the

attraction of outside resources, which may set off a strong cumulative process. The

in¯ ow of knowledge, as we have seen above, is driven both by actors from the

outside attracted into the region, and by local actors which try and tap into outside

knowledge.

Cumulative causation mechanisms might thus occur, giving rise to vicious and

virtuous circles which, again, strictly depend upon the sectoral points of strength

and weakness of both the MNC and indigenous ® rms.7 The `competitive bidding’

between European regional centresÐ in order to attract MNC’ s research and

innovation activitiesÐ seems to have become increasingly tough. In fact, the

bene® ts of the global generation of innovation are not likely to be evenly spread

between higher order and lower order regional locations. Furthermore, the risk of

regional inequalities might increase within national boundaries, as strong regional

systems of innovation would be further encouraged, while backward regions would

be further undermined by the strategies of MNCs.

As far as the latter point is concerned, we have stated above that location exerts

a strong centripetal pull on the innovative activities of MNCs. However, the

centrifugal forces which might offset the advantages of agglomerating in particular

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

390 J. Cantwell and S. Iammarino

regionsÐ such as rising prices of locally available resourcesÐ do not seem to have

a large in¯ uence on location decisions of this kind. The typical arguments for

convergenceÐ based on assumptions such as the price ± cost equalisation mechan-

ism, homogeneity of ® rms and sectors, positive incentives to locate in the

periphery, etc.Ð are clearly not applicable and it is hardly plausible that self-rein-

forcing regional growth (or decline) may be easily reversed by centrifugal forces

stemming from conventional market mechanisms. Technologically declining re-

gions, therefore, are not eligible as attractive locations for quality-seeking inward

investment, although, if the latter occurs, there may be some bene® ts in terms of

the revitalisation of past innovation capabilities through positive spill-over effects

(Cantwell, 1992a).

On the other hand, following the rationalisation of MNC innovation activity,

a stronger competition is likely to occur between higher order and lower order

locations, especially within an economically integrated area such as the EU. The

relationship between the internationalisation of innovation and the competitiveness

of local systems follows a circular pattern which Ð in the case of a virtuous

circle Ð goes from the upgrading of local innovatory capacity to the increase of the

share of research-related production, and from it to the improvement of export

performances, setting in motion a positive interaction between foreign and indige-

nous research and productive activity (Cantwell, 1992b). The role of MNC

networks for innovation is a function of the geographical hierarchy of centres, i.e.

of the strategic importance of the host region and of the innovative dynamism of

local competitors, suppliers and customers. The sectoral composition of techno-

logical strengths, in fact, differs across regional centres, while the technological

specialisation of foreign subsidiaries depends upon the rank of the regional location

in the geographical hierarchy. It has been pointed out that, by specialising

according to the local strength in each location, MNCs’ technological activity is

broadened, while the technological specialisation of regions, as already noted,

tends to become narrower (Cantwell, 1992b). Such interactions are more likely to

further upgrade higher order regional locations, in which the Local-for-local

strategy of subsidiaries, and their capacity to incorporate backward and forward

linkages in their external networks for innovation, aim at exploring local knowledge

and expertise, which will be integrated to widen technological competence at the

corporate level through the intra-® rm network (Local-for-global). Indeed, when

foreign research has a more pronounced explanatory nature, it is likely to be

attracted by higher order centres, treating them as a source of general expertise and

skills (Cantwell and Janne, 1997). On the other hand, lower order locations, with

a narrower scope of technological advantages, are seen as sources of speci® c

capabilities in some particular ® eld and thus they might be negatively affected due

to the Local-for-local strategy of foreign af® liates which follows a logic of exploi-

tation of indigenous expertise, with the possible aim of out-competing local rivals.

Furthermore, in the case of the EU, the globalisation of innovation through

MNC networks has been comparatively stronger than in other economic areas.

The removal of non-tariff barriers, the completion of the single European market

and the recent commitments to achieve the ultimate state of economic and

monetary integration have spurred the reorganisation of operations of MNC’ s

af® liates located in the EU to a much greater extent than in the case of af® liates

based outside the area. Cantwell (1992a) has shown that the degree of interdepen-

dency among geographically distinct units is relatively higher in Europe, where

inter-® rm networks and linkages between foreign af® liates and local ® rms turn out

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

M NCs, Innovation and Regional Systems 391

to be far more entrenched than in other areas of the world. The intra-area

rationalisation of economic activities boosted by the integration process has thus

given a strong impetus to the globalisation process, thereby strengthening the link

between Local-for-local and Local-for-global strategies and the effect by which

localisation economies lead to geographical agglomeration.

The effects of the globalisation of innovation by MNCs cannot thus be assessed

in general terms, as they depend principally upon the technological pro ® les and

strategies of MNCs, as well as upon the characteristics of regional systems of

innovation. However, despite the wealth of discussion on the changing division of

labour within MNCs across national borders, a paucity of both theory and

empirical evidence still exists on the bene® cial or detrimental interactions between

MNCs’ innovation activities and local systems of innovation.

The hypotheses to be tested here are that the regional location of innovative

activities by MNCs conforms to a geographical hierarchy of centres, and that the

composition of technological specialisation differs between those centres. Accord-

ingly, the technological specialisation of foreign-owned af® liates in each region

depends upon the position of that region in the locational hierarchy. In particular,

we hypothesise that as the position of the region in the hierarchy falls, so the pro ® le

of technological specialisation of foreign-owned ® rms in that region becomes more

closely related to the equivalent pattern of specialisation of indigenous ® rms in the

same region. Conversely, a centre at the top of the hierarchy is more likely to

attract a broad range of foreign innovative activities, as MNCs will generally try to

extend their established lines of specialisation through intra- ® rm networks. There-

fore, higher order locations should attract foreign research for their general

reservoir of skills and resources, while centres further down are attractive more for

their limited range of specialised expertise, thus bringing foreign and local techno-

logical pro® les closer together.

2. Regionalised Data and Methodology

From the above discussion, it becomes quite clear that foreign investment in

innovation has as much a regional scope as it has a national one. In particular,

recent trends in the EU support the conjecture that a comparative analysis at the

sub-national scale is the most appropriate way to identify the effects of globalisa-

tion. Therefore, we have tried to devise a more suitable and detailed geographical

unit of analysis, in order to throw some new light on the circumstances that give

rise to geographical agglomeration and increasing interactions between MNCs’

research location and regional systems of innovation.

The important issue for our purpose, actually, is not so much the precise

delimitation of the `local’ or `regional’ location in geographical terms, as to make

it clear that it is a geographical space characterised by a certain coherence based

on common behavioural practices linked to local institutions and culture (includ-

ing technological culture), industrial structure and corporate organisation

(Saxenian, 1994).

We decided to refer to sub-national entities singled out by normative criteria,

as classi® ed by the Eurostat in the Nomenclature of Territorial Units for Statistics

(NUTS), level 2, to provide a single uniform breakdown of territorial systems.

This classi ® cation is based on the institutional divisions currently in force in the

member states, according to the tasks allocated to territorial communities, to the

sizes of population necessary to carry out these tasks ef® ciently and economically,

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

392 J. Cantwell and S. Iammarino

and to historical, cultural and other factors. The NUTS 2 level (206 Basic

Regions) is generally used by the EU members for the application of their regional

policies and thus it is the most appropriate to analyse regional± national problems

(Eurostat, 1995).8

The empirical investigationÐ aimed at testing the relationship between the

foreign-owned and the indigenous company pro® les of technological specialisation

in regional centres ranked according to the geographical hierarchyÐ is carried out

by using patents granted to the world’ s largest industrial ® rms for innovation

located abroad, classi ® ed by the host European region in which the responsible

research facility is located.

The use of patents as an indicator of advanced technological capacity and

ability to develop innovation is one of the most established and reliable methods

of estimating innovative activities. The advantages and disadvantages of using

patent statistics are well known in the literature (see, among others, Pavitt, 1988;

Griliches, 1990; Archibugi, 1992) and will not be rehearsed here. For the purpose

of this paper, the use of patent records provides information on the address of both

the inventor and the owner of the invention (from which we have derived the

country of location of the parent ® rm through a consolidation of patents at the

level of international corporate group), thus allowing the identi® cation of where

the research and development underlying the invention was carried out in geo-

graphical terms. The regionalisation of our patent database consists of attributing

a revised, although still compatible, NUTS 2 code to each patent record, accord-

ing to the location of inventors in the EU countries, with reference to the period

1969± 95. Moreover, patents can be classi® ed by detailed technological ® elds

(grouped into 56 sectors in the database, see Appendix), which would not be

otherwise possible by using indicators such as, for example, R&D expenditure

(Zander, 1997). This is particularly appropriate here since the aim is to look at the

regional concentration of MNCs’ innovative activity in the EU, and to examine its

sectoral distribution at a sub-national level. Finally, the choice of US patenting is

convenient, since large ® rms are especially prone to patent their best quality

inventions in the US market, the largest and the most technologically advanced. It

is therefore likely that our data re¯ ect the patenting of inventions that have a

signi® cant commercial importance, as well as allowing for a meaningful analysis of

the territorial distribution of the technological operations of MNCs in the EU.

3. The Location of MNC Innovative Activities in the Italian Regions:

General Features

The regionalisation of the University of Reading patent database has brought to

light interesting insights on the geographical location of the innovative activities of

MNCs in the Italian case, which would not have been captured by the usual

analysis at the aggregate (country) level.

Con® rming the results of previous empirical studies on Italian technological

innovation at regional level (Cesaratto, Mangano and Silvani, 1993; Silvani et al.,

1993; Iammarino, Prisco and Silvani, 1998), the territorial distribution of inno-

vation generated in the af® liates of MNCs turns out to be highly polarised in just

two regionsÐ namely Lombardia and Piemonte. Therefore, we focused our atten-

tion on these two regional cores, for which the `Schumpeterian hypothesis’ seems

to hold on the basis that an average size of ® rm higher than for Italy as a whole

is associated with a relatively greater propensity to generate new technologies

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

M NCs, Innovation and Regional Systems 393

in-house and to commit to formal R&D activities (Iammarino et al., 1998). The

capacity to attract external resources shows that general external economies and

spillovers have been at work in the two regions, giving rise to the well-known

dichotomy of the Italian innovation system, i.e. the two regional cores on the one

hand, and the regional systems, such as Veneto and EmiliaÐ more oriented

towards engineering and design-based process innovations carried out by small

local ® rmsÐ where the degree of attraction of external research is very low.9

Table 1 reports the shares of US patents of the largest Italian ® rms attributable

to research located in Piemonte and Lombardia relative to that which they located

in Italy as a whole, and the equivalent regional shares for the research of

foreign-owned ® rms by technological sector in the period 1969± 95.10

As already emphasised, the geographical agglomeration of innovation is strik-

ing: more than 77% of the total research activity of large ® rms is concentrated in

these two regions. Foreign-owned research located in Italy appears to be relatively

more dispersed than that carried out by Italian ® rms: for foreign-owned ® rms the

aggregate share of the two regional cores is 68.4%, while for indigenous ® rms it is

as high as 82%. However, the capacity to attract externally-owned innovation

activity differs markedly between the two regional systems, supporting, at least at

® rst glance, our hypothesis of the existence of geographical hierarchies. While

Lombardia accounts for 57.1% of the total Italian-located research of foreign

subsidiaries (above the indigenous company share of 50.3%), the equivalent share

in Piemonte is only 11.3%, much lower that the regional share of patents granted

to the largest Italian ® rms (31.8%).

The sectoral distribution of patents demonstrates inter-regional differences to

an even greater extent. The key to the sectoral codes used in Table 1 is given in

the Appendix. In many sectors the large ® rm (Italian 1 foreign) total for the two

regions accounts for almost all the innovative activity carried out by such ® rms in

the country as a whole. However, in Lombardia the regional share of patents

attributable to the research of foreign ® rms is higher than the equivalent share for

large Italian ® rms in most sectors. In some sectors in particular Ð such as, for

example, Mechanical calculators and typewriters (30) or Of® ce equipment and

data processing systems (41)Ð the strong geographical agglomeration in the region

holds only for foreign af® liates (87.9% and 81.7% in these sectors respectively),

while the research carried out by Italian ® rms considered in the sample (respect-

ively 2.4% and 17.5% of the national total) has apparently gone mainly to other

regional locations. Conversely, the geographical concentration of foreign-owned

research in Piemonte is substantial and higher than that of Italian ® rms only in a

few sectors Ð namely Metal working equipment (17), Other general industrial

equipment (29), Internal combustion engines (42) and Rubber and plastic prod-

ucts (49). A noticeable case is that of Photographic chemistry (6), in which the

combined foreign share located in the two regions accounts only for 6% of total

foreign-owned corporate patents in this ® eld in Italy, and the local concentration

of foreign research is negligible in both cases.

Turning to the shares of total patents granted to large foreign-owned ® rms for

local research, Table 2 shows that the highest contribution of foreign research to

the regional total is once again recorded in Lombardia (39.1% for the period

1969± 95), con® rming the attractiveness of the region relative to both Piemonte,

with a foreign share of less than 17%, and to a lesser extent the country as a whole

(whose foreign share is 36.1%).

The sectoral picture again emerges as rather different between the two regional

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

394 J. Cantwell and S. Iammarino

Ta

ble

1.

Sh

are

so

fU

Sp

ate

nts

of

the

larg

est

Ital

ian

®rm

satt

rib

uta

ble

tore

searc

hin

Itali

an

regio

ns

rela

tive

toIt

aly

as

aw

ho

le,

an

dth

eeq

uiv

ale

nt

regio

nal

share

sfo

rth

eIt

ali

an

-lo

cate

dre

searc

ho

ffo

reig

n-o

wn

ed

®rm

s,b

yte

ch

no

logic

al

secto

r,1

96

9±9

5(%

)

Reg

ion

s

Pie

mo

nte

Lo

mb

ard

iaP

iem

on

te1

Lo

mb

ard

ia

Sec

tors

Itali

an®

rms

Fo

reig

n®

rms

To

tal

Itali

an

®rm

sF

ore

ign

®rm

sT

ota

lT

ota

l

33

6.5

11

.13

3.7

43

.22

2.2

41

.07

4.7

59

.46

.48

.26

8.6

66

.46

7.7

75

.8

66

6.7

1.7

3.4

33

.34

.35

.18

.5

72

6.7

7.1

21

.95

9.3

32

.15

2.6

74

.6

98

.52

.27

.56

6.7

38

.76

2.0

69

.5

11

15

.01

.41

0.6

67

.67

3.2

69

.48

0.0

12

3.3

2.2

2.9

80

.88

5.7

82

.78

5.6

13

50

.91

0.6

32

.42

1.8

27

.72

4.5

56

.9

14

41

.01

9.1

32

.13

7.0

45

.64

0.5

72

.6

16

17

.95

.91

4.8

49

.54

1.2

47

.36

2.1

17

33

.04

2.4

37

.25

0.5

38

.84

5.2

82

.4

18

14

.31

.93

.34

2.9

54

.75

3.3

56

.7

20

54

.81

1.4

39

.82

3.8

40

.92

9.7

69

.5

25

3.6

1.1

2.3

42

.99

6.7

71

.07

3.3

28

21

.31

8.9

20

.46

7.0

22

.65

1.0

71

.4

29

56

.85

7.4

57

.02

3.2

23

.82

3.4

80

.4

30

95

.91

2.1

82

.22

.48

7.9

16

.39

8.5

33

61

.87

.33

2.0

29

.48

0.5

57

.38

9.3

34

68

.86

.75

4.0

18

.86

0.0

28

.68

2.5

38

38

.82

0.1

26

.65

6.5

62

.66

0.5

87

.1

39

62

.34

.93

3.0

24

.66

8.8

47

.28

0.1

40

62

.51

.16

.33

7.5

73

.67

0.5

76

.8

41

71

.54

.83

9.5

17

.58

1.7

48

.38

7.8

42

54

.46

9.2

55

.63

2.7

15

.43

1.3

86

.9

49

9.2

16

.01

0.1

80

.42

0.0

71

.98

2.0

50

14

.36

.91

2.5

40

.77

5.9

49

.26

1.7

53

53

.77

.74

3.2

38

.45

1.6

41

.58

4.7

56

6.9

0.0

6.3

58

.64

0.0

57

.16

3.5

To

tal

56

31

.81

1.3

24

.45

0.3

57

.15

2.8

77

.2

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

M NCs, Innovation and Regional Systems 395

Table 2. Foreign shares (foreign-owned companies’percentage of total patents granted to large ® rms forlocal research in Italy), by sector and region, 1969 ± 95

Regions

Sectors Piemonte Lombardia Italy

3 3.6 5.9 10.8

5 31.8 40.1 40.9

6 50.0 83.3 97.5

7 8.0 15.0 24.6

9 4.9 10.6 16.9

11 4.1 33.8 32.0

12 29.4 39.6 38.3

13 15.2 52.0 46.1

14 24.1 45.6 40.5

16 10.3 22.4 25.8

17 51.4 38.8 45.2

18 50.0 90.6 88.3

20 9.8 47.4 34.4

25 25.0 71.2 52.3

28 33.3 16.0 36.1

29 34.9 35.3 34.7

30 2.4 87.9 16.3

33 12.5 76.7 54.7

34 2.9 50.0 23.8

38 49.6 67.7 65.4

39 7.5 74.4 51.1

40 16.7 95.5 91.6

41 5.8 81.1 47.9

42 10.1 4.0 8.1

49 22.2 3.9 14.0

50 13.3 37.3 24.2

53 4.1 28.5 22.9

56 0.0 5.6 7.9

Total 56 16.7 39.1 36.1

cores. Foreign shares of regional and national patents lower than the average are

found in some Chemicals, with the most notable exception of Photographic

chemistry (6), for which the contribution of foreign research to the sectoral

patenting activity is very high in all casesÐ 50% for Piemonte, 83.3% for Lombar-

dia and 97.5% for the country as a whole. It is useful to recall that, although the

foreign share in this sector is relatively high for both the regions considered (Table

2), the sector turned out to be the least concentrated in the two regional cores

(Table 1). Other examples of relatively small foreign shares in all three cases are

found in Chemical and allied equipment (16), Internal combustion engines (42),

Other instruments and controls (53) and Other manufacturing (56). On the

contrary, many of the sectors identi® ed as types of electrical equipment show

foreign shares much higher than the average both in Lombardia and in the country

as a wholeÐ for instance Semiconductors (40), for which research is due almost

entirely in both cases to af® liates of foreign multinationals. The only noteworthy

discrepancy between Italy and Lombardia is in Mechanical calculators and type-

writers (30), in which the regional patenting activity is attributable almost entirely

to foreign af® liates (88%), although foreign-owned ® rms represent a tiny pro-

portion of national patents in the sector (16.3%) and one of the lowest foreign

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

396 J. Cantwell and S. Iammarino

shares in Piemonte (2.4%). In the latter region, the sectors for which the shares of

research attributable to foreign ® rms register the highest values are generally to be

found in the broad group of mechanical engineering and machinery[e.g. Metal

working equipment (17), Paper making apparatus (18), Other specialised machin-

ery (28), Other general industrial equipment (29)] and in Electrical devices and

systems (38). The foreign share in Piemonte is high, relative to the low values for

both the country as a whole and Lombardia, also for Rubber and plastic products

(49).

Thus, it emerges that the sectoral distribution of foreign-owned companies’

shares of regional and national research activity shows a greater correspondence

between Lombardia and Italy, while in the case of Piemonte it seems to indicate

more focused sectoral patterns. It should be recalled that the narrower sectoral

spread of the foreign geographical concentration and shares of research in

Piemonte has to be understood also in relation to its high degree of productive

specialisation in some particular sectors in the broad industrial groups of mechan-

ical engineering and electrical equipment, for which the region has been labelled,

in relative terms, a mono-specialised region. On the other hand, Lombardia

appears to have a much wider sectoral spread of productive specialisation, as also

re¯ ected in the composition and competitiveness of regional exports (Iammarino

et al., 1998).

Finally, Table 3 reports the evolution over time of the foreign shares for the 56

sectors as a whole. It emerges clearly that the foreign proportion of total research

undertaken in Italy by large ® rms has increased markedly over the 27 years

considered here, reaching the highest values at the end of the period. This pattern

seems to be shaped by that of Lombardia, which has recorded a remarkable growth

of foreign innovative activities, particularly since the second half of the 1980s,

which had led by 1995 to a foreign share of 61.4%, three times bigger than that

in 1969. Conversely, the foreign share of regional patents in Piemonte shows a

much more unstable trend, actually ® nishing with a marginally lower share in the

® nal year than in 1969.

In general terms, the above picture seems to con® rm, at least in the Italian case,

that the location of technological activities of foreign-owned MNCs tends to be

strongly agglomerated at a sub-national level. This is evident particularly in

Lombardia, regardless of sectoral speci® cities, whereas in Piemonte the geograph-

ical concentration of foreign af® liates’ research seems to follow distinctive sectoral

patterns. Moreover, the strong attractiveness of Lombardia is demonstrated also

by its high foreign share of total regional patents, increasing over time more than

for the whole country. As argued above, the economic geography approach

suggests that the attractiveness of regional locations is determined by general

economies of scale, which lead to regional concentration, and by highly sector-

speci® c localisation economies. We can therefore establish, on the basis of a

different interaction between agglomeration forces in different regional systems, a

geographical hierarchy of regional centres in Italy. Lombardia can be considered,

from both the geographical and the sectoral perspectives, a higher order location

at the top of the scale, strongly attracting a broad range of foreign-owned

technological activities because of the general characteristics of its regional systems

of innovation. Piemonte comes next, with much lower values of both geographical

concentration within its boundaries and foreign shares of regional patenting and,

as a lower order location, it seems to be more attractive for some speci® c sectoral

features of the regional technological potential. The other Italian regions, although

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

M NCs, Innovation and Regional Systems 397

Table 3. Foreign shares (foreign-owned compa-nies’ percentage of total patents granted to large® rms for local research in Italy) by year and region,

1969± 95

Regions

Years Piemonte Lombardia Italy

1969 11.6 21.1 19.1

1970 15.2 23.5 23.7

1971 20.0 37.4 31.1

1972 18.9 29.4 29.0

1973 15.9 33.6 28.2

1974 12.1 27.5 26.6

1975 15.4 32.6 35.7

1976 7.7 31.3 28.7

1977 9.7 39.9 33.2

1978 12.3 32.3 28.6

1979 13.0 32.4 28.9

1980 16.9 31.9 31.9

1981 9.6 30.5 24.3

1982 11.7 21.4 19.3

1983 7.1 37.5 29.7

1984 29.6 35.3 33.2

1985 10.7 36.8 30.4

1986 24.4 42.4 37.4

1987 28.7 43.7 43.0

1988 18.3 29.6 34.2

1989 10.6 46.5 43.6

1990 14.5 63.6 52.1

1991 27.6 59.3 55.5

1992 18.2 58.3 54.0

1993 26.8 59.8 58.7

1994 18.8 57.3 58.6

1995 11.1 61.4 59.7

with some differences among them, are de® nitely behind, both in terms of

domestic innovative activity and even more in terms of the absolute level of

foreign-owned research that they are able to attract.

4. Technological Specialisation of MNCs and Regional Hierarchies

To investigate the characteristics of the pro® les of technological specialisation of

foreign-owned and local ® rms we have used the Revealed Technological Advan-

tage index (RTA), a proxy for the technological specialisation ® rst applied by Soete

(1987). The RTA index of a region in a particular technological sector is given by

its share of world patenting in that sector divided by its share of total world

patenting. Therefore, the RTA index is de® ned as follows:

RTA ij 5 (P i j / Pwj) ( S jP ij / S jPwj) where i 5 1, 211 and j 5 1, ¼ , 56 (1)

where P i j is the number of patents of region i in sector j and Pwj is the number of

world patents in the same sector. The index has been calculated for both Italian-

and foreign-owned ® rms in the two regional cores, Piemonte and Lombardia, and

in Italy as a whole for the period 1969± 95. It gives a measure of the performance

of (nationally-owned or foreign-owned) ® rms in a region in one particular sector

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

398 J. Cantwell and S. Iammarino

relative to their world performance. Since the RTA varies around unity, an index

greater than one suggests a comparative technological advantage, whilst an RTA

below unity indicates a position of comparative disadvantage.

Table 4 reports the RTA values by region and sector. In line with what has said

above, the technological advantages of both Italian and foreign-owned ® rms in

Piemonte, relative to Italian and foreign-owned ® rms in Lombardia and in Italy,

are more narrowly concentrated at sectoral level. It should be noted that, in sectors

such as Metal working equipment (17), Other general industrial equipment (29),

Mechanical calculators and typewriters (30) and Internal combustion engines

(42), for which local ® rms in Piemonte record a high specialisation both in

technology and in production, foreign af® liates also show a very strong geograph-

ical concentration of their research (see Table 1). This would support the

expectation that in Piemonte, de® ned as a lower order regional core, the attraction

of foreign resources is likely to be motivated by the willingness of MNCs to tap

into sectoral-speci® c local expertise and increase their own technological advantage

in some particular technological ® elds.

As far as Lombardia is concerned, the technological specialisation of Italian

and foreign-owned ® rms overlaps to a lower extent. In particular, Italian ® rms

show revealed comparative advantages mostly concentrated in the broad group of

chemicals and pharmaceuticals, in which foreign-owned research also appears to

have a technological advantage. As for Piemonte, in these sectors Ð such as

Chemical processes (5), Other organic compounds (11), Pharmaceuticals and

biotechnology (12) and also Textile and clothing machinery (25)Ð the geographi-

cal distribution of foreign technological activity is highly concentrated in the

region. On the other hand, foreign af® liates have an RTA greater than one also in

a number of sectors related to electrical equipmentÐ such as Electrical devices and

systems (38), Other general electrical equipment (39) and Semiconductors (40)Ð

and to of® ce equipment (30, 41); whereas the Italian large ® rms in the region

appear to be substantially despecialised. Foreign innovative activities in Lombar-

dia, although heavily concentrated in the region from a geographical perspective

and with a high contribution to regional innovation, seem therefore to coincide

with the technological expertise of local ® rms in only a limited number of sectors,

basically in chemicals and pharmaceuticals (5, 11 and 12) and in Textile and

clothing machinery (25), which represent points of strength also in the productive

and export specialisation pattern of local ® rms.

It is worth noting that, as far as Italy as a whole is concerned, the matching

between the technological specialisation of Italian ® rms and that of foreign

af® liates is relatively more pronounced, as in the case of Piemonte. The overlap is

particularly evident for Other organic compounds (11), Pharmaceuticals and

biotechnology (12), Metal working equipment (17) and in some sectors of

specialised industrial equipment (25, 28) and of® ce equipment (30). However, in

some electrical equipment sectors (38, 39, 40), the specialisation of foreign-owned

® rms does not correspond to an Italian technological advantage.

In order to test further our hypothesis that the technological specialisation of

foreign-owned subsidiaries in each region is associated with the position of the

region in the geographical hierarchy, a regression analysis was carried out. How-

ever, before going into the regression model, it is worth looking at the correlation

matrix between the RTA indices (Table 5), which provides some additional

insights on the characteristic features of the regional and national innovation

systems.

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

M NCs, Innovation and Regional Systems 399

Ta

ble

4.

RT

A(I

tali

an

an

dfo

reig

n®

rms)

rela

tive

toth

ew

orl

db

yse

cto

ran

dre

gio

ns,

19

69

±9

5

Reg

ion

s

Pie

mo

nte

Lo

mb

ard

iaIt

aly

Sec

tors

Ital

ian

®rm

sF

ore

ign

®rm

sIt

ali

an

®rm

sF

ore

ign

®rm

sIt

ali

an

®rm

sF

ore

ign

®rm

s

31

.82

80

.33

71

.37

10

.13

31

.59

60

.34

3

50

.26

40

.61

51

.21

61

.26

80

.89

31

.09

2

60

.06

40

.32

10

.02

00

.15

90

.03

12

.08

4

70

.68

80

.29

80

.96

50

.26

50

.81

90

.47

1

90

.48

10

.12

32

.37

80

.43

71

.79

30

.64

5

11

0.8

44

0.1

79

2.4

09

1.9

11

1.7

93

1.4

91

12

0.1

95

0.4

05

2.9

90

3.0

57

1.8

62

2.0

39

13

0.8

05

0.7

16

0.2

18

0.3

68

0.5

04

0.7

61

14

1.1

10

1.7

53

0.6

34

0.8

27

0.8

62

1.0

36

16

0.7

37

0.4

20

1.2

92

0.5

81

1.3

14

0.8

06

17

1.2

33

6.5

03

1.1

93

1.1

79

1.1

89

1.7

35

18

0.1

01

0.5

04

0.1

92

2.8

93

0.2

26

3.0

21

20

1.7

80

0.9

64

0.4

90

0.6

87

1.0

35

0.9

59

25

0.3

37

0.5

60

2.5

61

9.8

62

3.0

08

5.8

24

28

0.8

43

2.1

01

1.6

81

0.4

99

1.2

62

1.2

58

29

1.5

98

4.2

77

0.4

12

0.3

50

0.8

96

0.8

42

30

34

.56

14

.25

20

.54

06

.09

71

1.4

83

3.9

64

33

0.6

32

0.4

50

0.1

90

0.9

79

0.3

26

0.6

95

34

1.5

43

0.2

33

0.2

66

0.4

15

0.7

15

0.3

95

38

0.6

92

3.3

87

0.6

38

2.0

82

0.5

68

1.9

00

39

1.3

03

0.5

28

0.3

26

1.4

78

0.6

66

1.2

28

40

0.1

76

0.1

76

0.0

67

2.2

26

0.0

90

1.7

28

41

1.1

64

0.3

55

0.1

80

1.2

06

0.5

18

0.8

43

42

3.7

16

2.0

83

1.4

11

0.0

92

2.1

74

0.3

40

49

0.8

34

1.1

87

4.6

36

0.2

94

2.9

03

0.8

38

50

0.2

81

0.2

15

0.5

06

0.4

68

0.6

26

0.3

53

53

1.3

28

0.2

81

0.6

01

0.3

73

0.7

87

0.4

12

56

0.2

83

0.0

00

1.5

22

0.1

39

1.3

07

0.1

99

To

tal

Tech

56

11

11

11

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

400 J. Cantwell and S. Iammarino

Ta

ble

5.

RT

Aco

rrela

tio

nm

atr

ix

RT

AF

FR

TA

FF

LO

RT

AF

FP

IR

TA

ITF

RT

AIT

FL

OR

TA

ITF

PI

RT

AF

F1

0.8

39

0(0

.00

0)a

0.1

53

2(0

.26

0)

0.3

24

5(0

.01

5)b

0.0

06

8(0

.96

0)

0.3

00

0(0

.02

5)b

RT

AF

FL

O0

.83

90

(0.0

00

)a1

0.0

96

2(0

.48

1)

0.4

23

6(0

.00

1)a

0.1

00

2(0

.46

3)

0.3

45

6(0

.00

9)a

RT

AF

FP

I0

.15

32

(0.2

60

)0

.09

62

(0.4

81

)1

0.3

19

1(0

.01

7)b

20

.43

1(0

.75

3)

0.3

75

8(0

.00

4)a

RT

AIT

F0

.32

45

(0.0

15

)b0

.42

36

(0.0

01

)a0

.31

91

(0.0

17

)b1

0.3

81

9(0

.00

4)a

0.8

95

7(0

.00

0)a

RT

AIT

FL

O0

.00

68

(0.9

60

)0

.10

02

(0.4

63

)2

0.4

31

(0.7

53

)0

.38

19

(0.0

04

)a1

20

.04

44

(0.7

45

)

RT

AIT

FP

I0

.30

00

(0.0

25

)b0

.34

56

(0.0

09

)a0

.37

58

(0.0

04

)a0

.89

57

(0.0

00

)a2

0.0

44

4(0

.74

5)

1

56

case

s/2

tail

ed

sig

ni®

can

ce.

aS

ign

i®can

tat

1%

.b

Sig

ni®

can

tat

5%

.

RT

AIT

F:

RT

Ao

fIt

ali

an

®rm

sin

Italy

;R

TA

ITF

PI:

RT

Ao

fIt

ali

an

®rm

sin

Pie

mo

nte

;R

TA

ITF

LO

:R

TA

of

Itali

an

®rm

sin

Lo

mb

ard

ia;

RT

AF

F:

RT

Ao

fF

ore

ign

®rm

s

inIt

aly

;R

TA

FF

PI:

RT

Ao

fF

ore

ign

®rm

sin

Pie

mo

nte

;R

TA

FF

LO

:R

TA

of

Fo

reig

n®

rms

inL

om

bard

ia.

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

M NCs, Innovation and Regional Systems 401

The relevant point which emerges from Table 5 is that the specialisation

patterns of Italian ® rms in Lombardia and Italian ® rms in Piemonte are not

correlated, i.e. the regional hierarchy in Italy is associated with distinctive patterns

of regional specialisation. However, the Italian national system as a whole can be

understood as a composite of these two distinctive regional systems, as con® rmed

by the signi® cant correlation found between the RTA of Italian ® rms in the regions

and in the country as a whole (although the correlation coef® cient is much higher

for Piemonte than for Lombardia).

Accordingly, in the period 1969± 95, two geographical levels were taken into

consideration in the cross-section regression analysis: ® rst of all, the hypothesis was

tested at regional level for Piemonte and Lombardia, then for Italy as a whole. In

the former case, a log-linear model was assumed, since the distribution of the RTA

index for the two regions is skewed because of the small numbers of patents in

each of them, showing the existence of lognormality which is not the case for the

country as a whole.12 The regression was run across all the 56 technological sectors

(j).

lnRTAFFPI j 5 a 1 b lnRTAITFPI j 1 « j (2)

lnRTAFFLO j 5 a 1 b lnRTAITFLO j 1 « j (3)

RTAFF j 5 a 1 b RTAITF j 1 « j (4)

Furthermore, the period was broken down into 1969± 82 and 1983± 95 in order

to test the validity of the hypothesis at the two different geographical levels over

time.13 In this latter case we adopted a Granger notion of sequential causality, and

we run the regression for 47 technological sectors.14

lnRTAFFPIjt 5 a 1 b lnRTAITFPI jt 2 1 1 P jt (5)

lnRTAFFLO jt 5 a 1 b lnRTAITFLO jt 2 1 1 P jt (6)

RTAFF jt 5 a 1 b RTAITF jt 2 1 1 P jt (7)

where j refers to the sector of technological activity and t refers to the period

1983± 95 and t 2 1 to 1969± 82.

In the case of Piemonte (see Table 6), the pro® le of technological specialisation

of foreign-owned ® rms is closely related to the equivalent pattern of specialisation

of indigenous ® rms in the region. This supports our hypothesis that a lower order

region attracts ® rms more for its limited range of specialised expertise which

foreign research tries to tap into. This is also valid when looking at the regression

over time. The case of Lombardia is rather different (Table 7). Here, the

specialisation of foreign-owned ® rms does not depend on the technological advan-

tage of Italian ® rms, since there might be other factors bringing foreign ® rms into

the regionÐ such as, for example, spillovers due to the high concentration of

technological and productive activities, infrastructures, ® nancial facilities, etc. This

seems to suggest that a region at the top of the geographical hierarchy, such as

Lombardia, attracts a broad range of foreign-owned ® rms, generally extending

their lines of specialisation while drawing on the local source of general capabilities

and other characteristics of the regional innovation system. Finally, the combi-

nation of the two regional systems give us a picture of the overall Italian model

(Table 8), which is closer to the case of Piemonte (although, as expected, in

Piemonte the relationship between the technological advantages of foreign and

Italian ® rms is more robust than in the country as a whole), as the specialisation

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

402 J. Cantwell and S. Iammarino

Table 6. Results of the regressions for Piemonte

Results of the regression lnRTAFFPI j 5 a 1 b lnRTAITFPI j 1 « j

Coef® cient Standard error T-Ratio [Prob]

lnRTAITFPIj 0.5409 0.10928 4.9496 [0.000]a

Intercept 0.1500 0.09122 1.6446 [0.105]

R2

0.3120

LM Diagnostic Statisticsc

Absence of heteroscedasticity c 2 (DF 5 1) 0.58788 [0.443]No. of observations 56

Results of the regression lnRTAFFPI jt 5 a 1 b lnRTAITFPIjt 2 1 1 « jt

Coef® cient Standard error T-Ratio [Prob]

lnRTAITFPIjt 2 1 0.4049 0.1213 3.3299 [0.002]a

Intercept 0.2708 0.1094 2.4740 [0.017]b

R2 0.197

LM Diagnostic Statisticsc

Absence of heteroscedasticity c 2(DF 5 1) 0.01007 0 [0.920]

No. of observations 47

aSigni® cant at 1%.

b Signi® cant at 5%.c The LM test statistics reported are asymptotically distributed as a c 2 random variable;where c 2

(DF 5 1) 0.05 critical value: 3.84.

Table 7. Results of the regressions for Lombardia

Results of the regression lnRTAFFLO j, 5 a 1 b lnRTAITFLO j 1 « j

Coef® cient Standard error T-Ratio [Prob]

lnRTAITFLO j 0.11146 0.1585 0.7032 [0.484]Intercept 0.4294 0.1058 4.0580 [0.000]

a

R2

0.009

LM Diagnostic Statisticsc

Absence of heteroscedasticity c 2 (DF 5 1) 0.76238 [0.383]No. of observations 56

Results of the regression lnRTAFFLO jt 5 a 1 b lnRTAITFLO jt 2 1 1 « jt

Coef® cient Standard error T-Ratio[Prob]

lnRTAITFLO jt 2 1 0.1682 0.2030 0.8288 [0.411]Intercept 0.3954 0.1337 2.9558 [0.004]a

R2 0.197

LM Diagnostic Statisticsc

Absence of heteroscedasticity c 2(DF 5 1) 2.4508 [0.117]

No. of observations 47

aSigni® cant at 1%.

b Signi® cant at 5%.c

The LM test statistics reported are asymptotically distributed as a c 2random variable;

where c 2(DF 5 1) 0.05 critical value: 3.84.

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

M NCs, Innovation and Regional Systems 403

Table 8. Results of the regressions for Italy

Results of the regression RTAFF j 5 a 1 b RTAITF j 1 « j

Coef ® cient Standard error T-Ratio [Prob]

RTAITF j 0.2268 0.09000 2.5207 [0.015]b

Intercept 0.8786 0.17322 5.0724 [0.000]a

R2 0.1052

LM Diagnostic Statisticsc

Absence of heteroscedasticity c 2 (DF 5 1) 0.049296 [0.824]

No. of observations 56

Results of the regression RTAFF jt 5 a 1 b RTAITF jt 2 1 1 « jt

Coef ® cient Standard error T-Ratio [Prob]

RTAITF jt 2 1 0.2578 0.0899 2.8679 [0.006]a

Intercept 0.7155 0.1917 3.7318 [0.000]a

R2

0.154

LM Diagnostic Statisticsc

Absence of heteroscedasticity c 2 (DF 5 1) 0.2599E-4 [0.996]

No. of observations 47

aSigni® cant at 1%.

b Signi® cant at 5%.c

The LM test statistics reported are asymptotically distributed as a c 2random variable;

where c 2 (DF 5 1) 0.05 critical value: 3.84.

of the foreign and Italian ® rms matches for the overall period and even more

through time.

5. Conclusions

In this paper we have investigated the relationships between foreign MNCs and

regional systems of innovation in the Italian case. It has been possible, through the

regionalisation of the University of Reading database, to demonstrate the import-

ance of adopting a unit of analysis which allows for investigation below the

traditional country-level. This is essential in order to achieve a better understand-

ing of the details of the globalisation of technology and the organisation of

innovative activity by MNCs, and their implications for host locations.

The technological globalisation of multinational corporations, which was found

to conform closely to a geographical hierarchy of regional centres, might give rise

to a tougher competitive bidding between higher order and lower order regional

locations across European countries. This makes it increasingly important also to

consider more carefully the distribution of bene® ts and costs of integration

processes, and the potential gaps between the private and social returns to

innovative activity. Therefore, the region as a `unit’ can help determine the policies

implemented at the national or European level, in an obvious reversal of the

top/down relationship typically followed by public institutions.

For this purpose, our future work will be directed at improving the exercise

reported here, which needs to be extended at the European level. In fact, the

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

404 J. Cantwell and S. Iammarino

regionalisation of the equivalent data for other European locations, still in prog-

ress, will allow for comparisons so as to distinguish the sectoral patterns of

domestic and foreign technological specialisation in other European regions, and

to describe more appropriately the crucial link between the `global’ and `local’ .

Notes

1. The term `globalisation’ refers to a wider dimension than the one evoked by pre® xes to the word

national such as inter-, multi- or trans-. More speci® cally, by `globalisation’ we refer to a high

degree of interdependency among units which constitute the MNC. In principle, therefore, we

could have a higher interrelatedness among geographically dispersed units even with the same level

of internationalisation of innovative activities of the MNC.

2. Among those who claim that globalisation has largely in¯ uenced the bulk of economic and social

life, see Ohmae (1990) , Chesnais (1994) , Perraton et al., (1997) ; for a most skeptical view on its

actual quantitative relevance, see Ruigrok and van Tulder (1995) , Michie and Grieve Smith

(1995), Hirst and Thompson (1996) .

3. See, for instance, Cantwell (1995 ) and Patel (1995) . For an interpretation of the two sets of results

see Archibugi and Michie (1997) .

4. Among the studies in regional economics aimed at identifying the endogenous elements of

`territorialised’ innovative systems it is necessary to recall the approaches based on the concepts of

the milieux innovateur (Aydalot, 1986) and the technological district (Becattini, 1987). More

recently, attention has focused on speci® cally de® ned regional systems of innovation (Storper,

1995; Howells, 1999). As a consequence of the echoes of the ª new economic geographyº

(Krugman, 1991a, b; Grossman and Helpman, 1991), empirical analyses have also proliferated in

relation to the geography of innovation (among others, Jaffe, Trajtenberg and Henderson, 1993 ;

Audretsch and Feldman, 1994 , 1995 ; Breschi, 1997).

5. The distinction between these two types of agglomeration economiesÐ investigated over a long

period of time (from Marshall to Krugman, to mention but two)Ð is obviously not clear-cut. We

believe, however, that it turns out to be central in the assessment of the forces at work in analysing

the geographical pattern of MNC networks for innovation, as will be shown below.

6. An increasing number of studies have recently addressed this issue (see, for example, Neven and

Gouyette, 1994 ; Quah, 1996; Fagerberg and Verspagen, 1996; Fagerberg, Verspagen and CanieÈ ls,

1997). It turns out that the process of convergence in GDP across the EU regions, which was

observed during most of the post-war period up to the 1970s, is far from stable and, even

accounting for differences in industrial structure, it tended to slow down in the later part of the

1980s. By considering the differences in innovative capabilities across European regionsÐ even

more pronounced than at country levelÐ it has been shown that they account for a good deal in

explaining the diverging trends in economic growth (Fagerberg and Verspagen, 1996).

7. The idea that cumulative causation may give rise to a widening of regional inequalities is not a new

one: it goes back to Young (1928), Perroux (1950), Myrdal (1957) and Kaldor (1970 , 1981), in

its broader formulation within development economics. Only recently, however, it has received

renewed attention by some scholars interested in giving a fuller account of its interdependence with

the process of economic integration and the dynamics of globalisation (Cantwell, 1987 , 1989 ;

Cantwell and Dunning, 1991).

8. It should be noted, however, that despite the aim of comparing regions of equivalent size among

EU countries, this level still implies considerable differences between regions in terms of their area,

population, economic weight and administrative powers.

9. Piemonte and Lombardia can be labelled as regional cores for innovation also on the basis of other

indicators available. In fact, as the survey on technological innovation in the Italian manufacturing

industry has pointed out, 44% of all Italian innovative ® rms (7553, as reported by the survey) are

located in the two regions, and more than 50% of total national costs of innovation is spent in

Lombardia and Piemonte. The latter percentage grows even moreÐ above 60%Ð when considering

the formal R&D expenditure (Iammarino et al., 1998).

10. Some of the technological sectors were dropped from the Tables on the grounds of the relatively

small number of patents for corporate research in the country as a whole, the outcome being that

only 28 sectors are reported in Tables 1± 4 and commented in the text. The total 56, however,

refers to the total number of patents for all 56 technological sectors.

11. For Italy as a whole, the index obviously considers S i, where i 5 1, ¼ , 20 (the 20 Italian regions).

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

M NCs, Innovation and Regional Systems 405

12. The standard assumption of this methodology is that the regression is linear, which is valid if the

cross-sector index approximately conforms to a normal distribution. However, as discussed more

extensively in Cantwell (1991) , if the sample of patents numbers is really insuf® ciently large, an

arti® cially high degree of dispersion in the index is very likely to be found, as is the case of the two

regions here considered. This is what tends to lead to a skewed distribution, rather than any

property of the RTA index itself. Thus, for the two regions, the sectoral distribution represented

by the RTA conforms more closely to a lognormal than a normal distribution, so taking a

logarithmic rather than a linear functional form in such cases improves the values of the t-statistic

on b .

13. Arguably, according to the concept of globalisation of innovation as de® ned in Section 1, such a

process is likely to have taken off especially from the 1980s onwards. Thus, we might expect our

hypothesised relationship to become stronger over the period (i.e. 1969 ± 95).

14. The reason for having a smaller number of sectors in the lagged cross-section model is that, when

we subdivided the period 1969 ± 95, we dropped all the technological sectors with an overall

number of patents less than 600 in the world total in both 1969 ± 82 and 1983 ± 95. The purpose was

to avoid the inclusion of sectors with a relatively low propensity to patent at the world level.

References

Amin, A. & Tomaney, J., ª The Regional Development Potential of Inward Investment in the Less

Favoured Regions of the European Community,º in A. Amin and J. Tomaney, eds, Behind the Myth

of the European Union, London: Routledge, 1995.

Archibugi, D., ª Patenting as an Indicator of Technological Innovation: A Review,º Science and Public

Policy, 1992, 19, 357± 68.

Archibugi, D. & Iammarino, S., ª Innovation and Globalisation: Evidence and Implications,º Italian

National Research Council Working Paper, July 1997.

Archibugi, D. & Iammarino, S., ª The Policy Implications of the Globalisation of Innovation,º ESRC

Centre for Business Research Working Paper, University of Cambridge, July 1997, No. 75.

Archibugi, D. & Michie, J., `Technological globalisation and national systems of innovation: an

introduction’ , in D. Archibugi and J. Michie (eds), Technology, Globalisation and Economic Perform-

ance, Cambridge: Cambridge University Press, 1997.

Audretsch, D.B. & Feldman, M., ª Knowledge Spillovers and the Geography of Innovation and

Production,º CEPR Discussion Paper No. 953, 1994.

Audretsch, D.B. & Feldman, M., ª Innovative Clusters and the Industry Life Cycle,º CEPR Discussion

Paper No. 1161, 1995.

Aydalot, P., ed., Milieux innovateurs in Europe, Paris: Gremi, 1986.

Becattini, G., Mercato e forze locali: il distretto industriale, Bologna: Il Mulino, 1987.

Breschi, S., ª The Geography of Innovation: a Cross-Sector Analysis,º CESPRI Working Paper No. 95,

1997.

Cantwell, J.A., ª The Reorganization of European Industries after Integration: Selected Evidence on the

Role of Multinational Enterprise Activities,º Journal of Common Market Studies, December 1987 ,

26(2).

Cantwell, J.A., Technological Innovation and the Multinational Corporation. Oxford: Basil Blackwell,

1989.

Cantwell, J.A., ª Historical Trends in International Patterns of Technological Innovationº , in J.

Foreman-Peck, ed., New Perspectives on the Late Victorian Economy. Cambridge: Cambridge Univer-

sity Press, 1991.

Cantwell, J.A., ª The Effects of Integration on the Structure of Multinational Corporation Activity in

the EC,º in M.W. Klein and P.J.J. Welfens, eds, Multinationals in the New Europe and Global Trade.

Berlin: Springer, 1992a.

Cantwell, J.A. ª The Internationalisation of Technological Activity and its Implications for Competitive-

ness,º in O. Granstrand, L. HaÊ kanson and S. SjoÈ lander, eds., Technology Management and

International Business. Internationalization of R&D and Technology. Chichester: Wiley, 1992b .

Cantwell, J.A., ed., Transnational Corporation and Innovatory Activities. London: Routledge, 1994 .

Cantwell, J.A., ª The Globalisation of Technology: What Remains of the Product Cycle Model?,º

Cambridge Journal of Economics, 1995, 19(1), pp. 155± 74.

Cantwell, J.A. & Dunning, J.H., ª MNEs, Technology and the Competitiveness of European Indus-

tries,º Aussenwirtschaft, 1991, 46, pp. 45± 65.

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

406 J. Cantwell and S. Iammarino

Cantwell, J.A. & Janne, O., ª Technological Globalisation and Innovative Centres: the Role of

Corporate Technological Leadership and Locational Hierarchy,º Discussion Papers in International

Investment & Management, University of Reading, No. 239, 1997.

Cesaratto, S., Mangano, S. & Silvani, A. ª L’ innovazione nell’ industria italiana secondo l’ indagine

Istat-CNR: tipologie settoriali e dinamiche territoriali,º Economia e politica industriale, 1993, 79,

pp. 167 ± 200.

Chesnais, F., ª Technical Co-operation Agreements Between Firms,º Science Technology Industry

Review , 1988, 4, 57 ± 119.

Chesnais, F., La mondialisation du capital, Paris: Syros, 1994.

DeBresson, C., ª I poli tecnologici dello sviluppo,º L’ industria, 1987 , 3, pp. 301 ± 35.

Dunning, J. & Robson, P., ª Multinational Corporate Integration and Regional Economic Integration,º

Journal of Common Market Studies, 1987, 24(2).

Dunning, J.H., Multinational Enterprises and the Global Economy. Workingham: Addison-Wesley, 1993 .

Dunning, J.H., ª Multinational Enterprises and the Globalisation of Innovatory Capacity,º Research

Policy, 1994, 23, pp. 67± 88.

Dunning, J.H. & Wymbs, C., ª The Geographical Sourcing of Technology Based Assets by Multina-

tional Enterprisesº , Rutgers University, mimeo, 1997 .

Eurostat, NUTS. Nomenclature of Territorial Units for Statistics, Luxembourg, 1995.

Fagerberg, J. & Verspagen, B., ª Heading for Divergence? Regional Growth in Europe Reconsidered,º

Journal of Common Market Studies, 1996, 34(3), pp. 431± 48.

Fagerberg, J., Verspagen, B. & CanieÈ ls, M., ª Technology Gaps, Growth and Unemployment Across

European Regions,º MERIT, University of Maastricht, mimeo, 1997.

Ghoshal, S. & Barlett, C.A., ª Innovation Processes in Multinational Corporations,º Strategic Manage-

ment Journal, 1990, 11, pp. 499± 518.

Griliches, Z., ª Patent Statistics as Economic Indicators: A Survey,º Journal of Economic Literature,

1990 , 28, pp. 1661 ± 707.

Grossman, G.M. & Helpman, E., Innovation and Growth in The Global Economy, Cambridge, MA: MIT

Press, 1991.

HaÈ gerstrand, T., Innovation Diffusion as a Spatial Process, Chicago: University of Chicago Press, 1967 .

Hirst, P. & Thompson, G., Globalization in Question, Cambridge: Polity Press, 1996.

Howells, J., ª Regional Systems of Innovation?,º in D. Archibugi, J. Howells and J. Michie, eds,

Innovation in the Global Economy, Cambridge: Cambridge University Press, 1999 , forthcoming.

Howells, J. & Wood, M., The Globalisation of Production and Technology, London: Belhaven Press, 1993 .

Iammarino, S., Prisco, M.R. & Silvani, A., ª On the Importance of Regional Innovation Flows in the

European Union: Some Methodological Issues in the Italian Case,º Research Evaluation, 1995, 5(3),

pp. 189 ± 206.

Iammarino, S., Prisco, M.R. & Silvani, A., ª The Geography of Production and Innovation: How

Regional ª Stylesº Play in the Global Scenarioº , Review of Regional Science, 1998, forthcoming.

Jaffe, A.B., Trajtenberg, M. & Henderson, R., ª Geographic Localization of Knowledge Spillovers as

Evidenced by Patent Citations,º Quarterly Journal of Economics, 1993, 63(3), pp. 577± 98.

Kaldor, N., ª The Case for Regional Policies,º Scottish Journal of Political Economy, 1970, 17(4), pp.

337± 348.

Kaldor, N., ª The Role of Increasing Returns, Technical Progress and Cumulative Causation in the

Theory of International Trade and Economic Growth,º Economie applique e, 1981 , 34(4), pp. 593 ±

617.

Kline, G.J. & Rosenberg, N., ª An Overview of Innovationº , in R. Landau and N. Rosenberg, eds, The

Positive Sum Strategy: Harnessing Technology for Economic Growth, Washington, DC: National Acad-

emy Press, 1986.

Krugman, P., Geography and Trade, Cambridge, MA: MIT Press, 1991a.

Krugman, P., ª Increasing Returns and Economic Geography,º Journal of Political Economy, 1991b,

99(33), pp. 483± 99.

Lundvall, B.AÊ ., ª Innovation as an Interactive Process: From User± Producer Interaction to the National

System of Innovation,º in G. Dosi et al., eds, Technical Change and Economic Theory, London: Pinter,

1988.

Lundvall, B.AÊ ., ed., National Systems of Innovation, London: Pinter, 1992.

Malmberg, A., SoÈ lvell, O. & Zander, I., ª Spatial Clustering, Local Accumulation of Knowledge and

Firm Competitiveness,º Geogra® ska Annaler, 1996, 78B(2), pp. 85± 97.

Michie, J. & Grieve Smith, J., eds, Managing the Global Economy, Oxford: Oxford University Press,

1995.

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

M NCs, Innovation and Regional Systems 407

Miller, R., ª Global R&D Networks and Large-Scale Innovations: The Case of the Automobile

Industry,º Research Policy, 1994 , 23, pp. 27 ± 46.

Myrdal, G., Economic Theory and the Underveloped Regions, London: Ducksworth, 1957 .

Neven, D.J. & Gouyette, C., ª Regional Convergence in the European Community,º CEPR Discussion

Paper No. 914, 1994.

Ohmae, K., The Borderless World: Management Lessons in the New Logic of the Global Market Place,

London: Collins, 1990 .

Patel, P., ª Localised Production of Technology for Global Markets,º Cambridge Journal of Economics,

1995 , 19(1) pp. 141± 53.

Pavitt, K., ª International Patterns of Technological Accumulation,º in N. Hood and J. Vahlne, eds,

Strategies in Global Competition, London: Croom Helm, 1988.

Pearce, R.D. & Singh, S., Globalizing Research and Development, London: Macmillan, 1992 .

Perraton, J., Goldblatt, D., Held, D. & McGrew, A., ª The Globalisation of Economic Activity,º New

Political Economy, 1997, 2(2), pp. 257± 278.

Perroux, F., ª Economic Space: Theory and Applications,º Quarterly Journal of Economics, 1950, 64(1),

pp. 89± 104.

Quah, D.T., ª Regional Convergence Clusters Across Europe,º CEP± LSE Discussion Papers, No. 274,

February 1996.

Ruigrok, W. & van Tulder, R., The Logic of International Restructuring, London: Routledge, 1995 .

Saxenian, A., Regional Advantage. Culture and Competition in Silicon Valley and Route 128, Cambridge,

MA: Harvard University Press, 1994 .

Soete, L., ª The Impact of Technological Innovation on International Trade Patterns: The Evidence

Reconsidered,º Research Policy, 1987, 16, pp. 101 ± 30.

Silvani, A., De Bresson, C., Berni, A. and Hu, X., ª La localisation regionale des grappes d’ innovation

en Italie: troisieme Italie ou Lombardie?,º Revue d’Economie Regionale et Urbaine, 1993, 2, pp. 289 ±

307.

Storper, M., ª The Resurgence of Regional Economies, Ten Years Later: The Region as a Nexus of

Untraded Interdependencies,º European Urban and Regional Studies, 1995, 2, pp. 191± 221.

Young, A., ª Increasing Returns and Economic Progress,º Economic Journal, 1928, 38(152), pp. 527 ±

42.

von Hippel, E., The Sources of Innovation, Oxford: Oxford University Press, 1989.

Zander, I., ª How Do You Mean ª globalº ? A Taxonomy of Innovation Networks in the Multinational

Corporation,º Paper presented at the Seminar on the ª Internationalization of Corporate R&Dº ,

CIRST, Universite de Montre al, August 1997.

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014

408 J. Cantwell and S. Iammarino

Appendix: Sectoral groups

1 Food and tobacco products

2 Distillation processes

3 Inorganic chemicals

4 Agricultural chemicals

5 Chemical processes

6 Photographic chemistry

7 Cleaning agents and other compositions

8 Disinfecting and preserving

9 Synthetic resins and ® bres

10 Bleaching and dyeing

11 Other organic compounds

12 Pharmaceuticals and biotechnology

13 Metallurgical processes

14 Miscellaneous metal products

15 Food, drink and tobacco equipment

16 Chemical and allied equipment

17 Metal working equipment

18 Paper making apparatus

19 Building material processing equipment

20 Assembly and material handling equipment

21 Agricultural equipment

22 Other construction and excavating equipment

23 Mining equipment

24 Electrical lamp manufacturing

25 Textile and clothing machinery

26 Printing and publishing machinery

27 Woodworking tools and machinery

28 Other specialised machinery

29 Other general industrial equipment

30 Mechanical calculators and typewriters

31 Power plants

32 Nuclear reactors

33 Telecommunications

34 Other electrical communication systems

35 Special radio systems

36 Image and sound equipment

37 Illumination devices

38 Electrical devices and systems

39 Other general electrical equipment

40 Semiconductors

41 Of® ce equipment and data processing systems

42 Internal combustion engines

43 Motor vehicles

44 Aircraft

45 Ships and marine propulsion

46 Railways and railway equipment

47 Other transport equipment

48 Textiles, clothing and leather

49 Rubber and plastic products

50 Non-metallic mineral products

51 Coal and petroleum products

52 Photographic equipment

53 Other instruments and controls

54 Wood products

55 Explosive compositions and charges

56 Other manufacturing and non-industrial

Dow

nloa

ded

by [

Nip

issi

ng U

nive

rsity

] at

05:

46 0

6 O

ctob

er 2

014