Local Cluster Concepts, Facts and Theory The Topic Concepts Theory Facts Thomas Brenner DIMETIC October 2007.

Local ClusterConcepts, Facts and Theory

The Topic Concepts Theory Facts

Thomas Brenner

DIMETIC October 2007

Why Is This Topic Relevant ?

Scientific perspective: Spatial distribution of industries Success of certain regions

Policy perspective: Why are there differences between regions ? How can unsuccessful regions be made

successful ? How can clusters be established ? How can they be made more successful ?

How Is This Topic Treated in the Literature ?

Case studies: Study of emergence Examination of success

Theoretical approach: Mathematical: New Economic Geography Conceptual: Various concepts

Spatial econometrics: Spatial distribution of industries Identification of clusters

New Economic Geography

Mathematical Two- or few-country models of specialisation Stability of local clusters

Simulations Local externalities: centripedal and

centrifugal forces Explanation for existence of agglomerations Spatial distribution

Conceptual Works

Discussion of one or a few important factors

Industrial districts Innovative milieux Local clusters

Various definitions Dominant definition: Porter

Learning regions, regional innovation systems etc.

Discussion of Important Factors

Common labour market, service firms and cooperation (Marshall)

Flexible specialisation (Piore & Sabel) Social networks (Becattini) Diamond model (Porter) Innovative milieux (Camagni) Learning regions (Lundvall) Spin-offs (Klepper)

Industrial Districts

Marshallian: Common labour market Service firms Spillover / Interaction

Italian: Plus: Social network

Characteristics Concentration of small firms Strong interaction Manufacturing industry

Innovative Milieux

Concentration of innovative actors This creates sometimes

Synergies through interaction Innovative atmosphere

Porter’s Diamond Model

Chance

policy

Local factors and resources

Firm strategy and competition

Related and service industries

Demand

Definition: What Are Clusters ?

Scientific: Many different concepts and definitions Common: Agglomeration of firms of one or a

few industries in a region Different additional conditions, industry and

region requirements

Policy: Some regionally interacting firms Sometimes higher requirements

Factors: What Causes/Characterises Clusters ?

Supplier-buyer linkages (ID,IM,LC) Cooperation (ID,IM) and competition (LC) Social contacts (ID) Innovations (IM) Local research (LC) Local policy (LC) Spillover/synergies (IM) Labour market (ID) Demand (LC)

Emergence vs. Success

Success How do firms benefit in clusters ? What are the mechanisms behind local

externalities ? Does not explain existence !

Emergence Why do local clusters exist ? How do they emerge ? What are the mechanisms behind the self-

reinforcement of clusters ?

positive feedback-loops:

example:

Theory: Self-augmenting Processes

Number of firms

Founding situation

Mechanisms (examples):• Accumulation of human capital• Cooperation among firms• Choice of co-location• Spillover• Interaction with public research• Spinoffs• Interaction with local policy

Theoretical Model (Schematic)

External conditions and fixed local factors

Local firm population

Flexible local conditions

Endogenous dynamics

Theoretical Model (Mathematical)

e

f l

fflfff tfrtlatfatfefadt

tdffflfffef

)()()())()(ˆ()(

llfl tlrtfadt

tdlllfl

)()()(

Stable States I

Weak local self-augmenting processes:

e

f

Stable States II

Weak local self-augmenting processes:

e

f

Stable states

Unstable states

Two stable states:

Prediction: Static situation

Number of firms

Regions with cluster

Regions without cluster

Critical mass

Number of regions

Prediction: Distribution

Should hold for• Number of firms• Number of employees• Number of patents

Number of regions

Number

critical mass

Example: Printing

0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035

Anteil an der gesamten Beschäftigtenzahl

0

1

2

3

4

log(

Häu

figke

it)

Number of employees

Log (frequency)

Example: Clothing

0.00 0.01 0.02 0.03 0.04 0.05 0.06


0

1

2

3

4

5

log(

Häu

figke

it)

Number of employees

Log (frequency)

Example: Glas

0.00 0.02 0.04 0.06 0.08 0.10 0.12


0

1

2

3

4

5

log(

Häu

figke

it)

Number of employees

Log (frequency)

Emergence of Clusters

What are the relevant dynamics and mechanisms ?

What are the relevant local factors ? How do we study these processes ?

How do local clusters emerge?

1st stage: Emergence of the market



2nd stage: Emergence of cluster




3rd stage: Stability




3rd stage: Stability

4th stage: Disappearance

Necessary conditions I

I. Sufficiently strong local self-augmenting processes

Necessary conditions II

1st stage: Emergence of the market:II. Sufficiently supportive local conditions

Market and local attractiveness

Necessary conditions III

2nd stage: Emergence of cluster:III. Sufficiently fast development


Necessary conditions

Substitutes: Different mechanisms creating self-augmenting

dynamics Different aspects of local conditions Different aspects of local developments

Complements: The self-augmenting dynamics have to be strong

enoughand

The local conditions have to be sufficiently good and

The local developments have to be sufficiently fast

Analysis of Factors and Dynamics

Case studies What were the initial conditions ? How did the region develop ?

Meta-study: Van der Linde 2003 (Internet, Diamond

model) Own study: 159 cases Analysis of the literature (222 works)

Meta-study

Factors, dynamics and events analysed 12 self-augmenting processes 17 local factors 6 events

Characterisation: I: is argued to be important and present U: is argued to be unimportant or ot present N: is not mentioned

Self-augmenting Processes

Process I U N

F-HC 116 10 33

COOP 87 22 50

COLOC 83 3 73

INTRA-SPILL 81 14 64

F-EDU/RES 66 19 74

SPIN 60 4 95

F-POL 49 10 100

INTER-SPILL 46 1 112

F-OPIN 44 9 106

F-VC 35 6 118

Local Factors

Factor I U N

LABOUR 105 10 44

NET 78 37 44

UNI/RES 70 22 67

TRAD 66 10 83

IND 61 2 96

LOC-POL 56 18 85

INFRA 52 10 97

CULT 52 14 93

GEOGR 51 2 106

DEMAND 49 20 90

Events

Dynamics I U N

LEAD-FIRM 62 4 93

SPEC-POL 53 10 96

HIST 52 0 107

PROM 22 7 130

INNO 15 4 140

Comparison: Time

Mechanism Post 1950 vs. Ante 1950

Post 1970 vs. Ante 1950

F-EDU/RES >

F-VC >

LOC-POL >

GEOGR <

TECH-PARK > >

IND <

SPEC-POL > >

CHANCE <

Comparison: Industry

Mechanism High tech vs. Low

tech

Know. vs. Non-know.

SPIN, SUPSTART, F-EDU/RES > >

F-HC, F-VC >

COOP < <

NAT-POL, UNI/RES, LABOUR, TECH-PARK, CAPITAL, LIFE

> >

LOC-POL, TYPE >

GEOGR, WAGE < <

SPEC-POL, LEAD-FIRM > >

HIST <

Comparison: Country

Mechanism Developed vs. developing

Anglo-saxion

Asia

F-EDU/RES, SPIN > <

F-POL, F-OPIN <

COOP >

WAGE <

LIFE > <

CULT, UNI/RES <

TYPE, INFRA >

LEAD-FIRM > < <

Measuring Regional Innovation Systems

What is a RIS? Innovativeness of regions Measurement approaches Local factors

Thomas Brenner


What is a RIS ?

Two definitions in the literature: Innovative regions

Research infrastructure Innovative firms and industries

Similar to national innovation systems All elements and interactions That contribute to innovation generation In a region

What is a RIS ?

Innovative regions How are they defined?

Case studies Innovation output, but no threshold Innovation inputs, such as R&D expenditures

or public research

What does it mean?

What is a RIS ?

System perspective Number of interacting elements With interactions like

Joint research Cooperation (more general) Flow of knowledge Flow of people Flow of money

Innovativeness of Regions

Regional Innovation Scoreboard Ranking of EU regions according to a number

of variables Human resources in science and technology Participation in life-long learning Public R&D expenditures Business R&D expenditures Employment in medium/high-tech manufacturing Employment in high-tech services European patent applications


From a presentation by Hugo Hollanders (MERIT)


What is innovativeness of regions? Total innovation output? Innovation output per inhabitant? Innovation output per R&D employee?

What does it mean?


Measurement Approaches

Usual approaches: Innovation scoreboard

Input and output variables

Total innovation output Innovation output per inhabitant Innovation output per R&D employee Knowledge production function and

production frontier Non-parametric frontier approach

Local Factors

What are the local factors that play a role for the innovation output of regions: Regression

Nations Regions Firms

Questionnaire How to frame the questions ?

Case studies How to generalise ?

Local Factors

What are the local factors that play a role for the innovation output of regions

What are the implications for policy? What are the implications for firms?

FE PR HK NW OI CU PO LM KP

Identification of Local Clusterand Policy Issues

Identification of Local Clusters Problems and Local Resources Policy activities Policy recommendations

Thomas Brenner


Identification of Local Clustering

Which Industries show clustering ? Ignored in the literature Some discussion about low-tech industries

Where are local clusters ? Italian approach Identification according to specialisation ratio Own approach: Distribution estimation

Italian Approach

Definition of labour market areas Specialisation (share of industry’s workers) Additional conditions

Dominance of manufacturing Dominance of small and medium size firms

Results Identification of industrial districts Repeated

Specialisation ratio

Regions according to statistical office Specialisation (share of industry’s workers)

higher than 3 Sometimes additional conditions

Certain number of related industries

Results Identification of local clusters

Distribution Estimation

Exponential function:P(f) = a1

f

Boltzmann-like distribution:P(f) = f a2

f

Cluster part:P(f) = (f-a4) a5

(f-a4) if f>a4

Industries with Local Clusters

Manufacturing industries: Clustering can be proved for around 50%

For the number of workers and For the number of firm sites

Service industries: Clustering can be proved for around 30% for

each number Approach does not apply for around 20%


sub-industries industries with clusters

food 9 3

tabacco 1 -

textile 7 6

clothing 3 2

leather 3 3

wood 5 4

paper 5 -

printing 3 2

petroleum processing 3 1

chemicals 7 1

plastics 2 1

glas, ceramics & stones 8 4


sub-industries industries with clusters

metal production 5 5

metal products 7 4

machinery 7 2

office machines 1 -

electronics 6 2

telecommunication 3 2

instruments 5 3

cars 3 -

other transportation 5 3

furniture, toys, ... 6 4

recycling 2 2

Schleswig-Holstein

Hamburg

Niedersachsen

Bremen

Nordrhein-Westfalen

Hessen

Rheinland-Pfalz

Baden-Württemberg

Bayern

Saarland

Berlin

Brandenburg

Mecklenburg-Vorpommern

Sachsen

Sachsen-Anhalt

Thüringen

Problems with Identification

Local rescourcesFrequency

Number of employees

Frequency


Number of employees

Local rescources

Frequency

Frequency


What are the reasons for industrial agglomerations?

Local externalities Local resources History

Approaches Ellison/Glaeser: Dartboard approach Bottazzi et. al.: Model of local externalities Own approach: Model of local externalities

plus local resources

Location Model

kkksentryklacnp exp)(

Entry:

kkkxexitklbncnp exp)(

Exit:

Example: Automobiles

Number of firms

Local resources

Problems with Location Model

Not all factors can be empirically measured What are the original factors ? What factors develop because of the

agglomeration of the industry ?

Policy approaches

Cluster organisation Network financing and promotion Monetary support for cluster activities Technology parks Improvement of local conditions

What is supported

Coordination Networking Joint innovation and research Subsidising specific locations Public research Education

Basic questions

Should policy be involved ? When should activities been taken ? Where should activities been taken ? What should be done ?

First Stage: Where Do Clusters Emerge?

1st and 2nd stage: Emergence of clusters:Sufficiently supportive local conditions


Second Stage: Will Potential be Used ?

2st stage:improve local dynamics


Third Stage: Stable Situation

3rd stage:Emergence unlikely


Which is the Right Region ?

Should locations be supported by policy?

Number of regions

Activity

critical mass

• NO

• YES• NO

Contact:

[email protected]




Knowledge Production Function

Cobb-Douglas function

Inputs are R&D expenditures Universities Local factors (such as population density,

GDP, education, …)

Linear function

i

iiInputaoutputInnovation

i

iiInputaoutputInnovation


Non-parametric frontier approach

output

input

x x

x

x

x

xx

x

xx

xx

x

x

x

xx

x


Private R&D

Mechanisms Firms’ R&D employees produce most

innovations

Empirical evidence High correlation R&D expenditure is even often used to

measure innovativeness

Implications Innovativeness depends on innovative firms

Public Research

Mechanisms Source of ideas and knowledge Cooperation partner Source of human capital

Empirical evidence Econometric evidence, regionally bound Source of ideas: average importance Cooperation: 13%

Implications Firm location Set-up of research institutes

Education

Mechanisms Availability of highly qualified labour

Empirical evidence Evidence for university graduates Importance for large firms

Implications Firm location Set-up of universities

Education

Small firms

Medium firms

Large firms

With R&D

Without R&D

Qualified labour 2,22 2,52 3,26 3,05 2,18

Ideas for new prod.&proc. 2,53 2,4 2,04 2,19 2,5

Information 2,32 2,24 2,33 2,29 2,31

Direct support 2,08 2,28 2,67 2,37 2,25

New instruments and techniques

2,22 2,20 2,42 2,39 2,17

Results from basic research 2,19 2,04 1,96 2,18 1,98

Consulting 1,86 1,96 2,33 2,02 2,04

Relevance of Universities:

Networks and Cooperation

Mechanisms Most innovations are produced jointly 13% are done allone

Empirical evidence Cooperation is frequent Effect is not proved

Implications ??

Networks and Cooperation

Partner Percentage

Material and component supplier 55,2%

Machinery supplier 54%

Customer 43,7%

Associations 35,1%

Other firm sites of the same company 27,9%

Universities 25,9%

Central lab of the company 24,2%

Public research institutes 21,5%

Consulting firms 19,3%

Competitors 8,5%

Capital market

Mechanisms Financial resources for innovations

Empirical evidence High percentage for start-ups and first

innovation project, than decreasing Local effect ?

Implications Start-up process important

Labour market laws

Mechanisms Flexibility important for innovation projects Innovative employees need liberty and

belongingness

Empirical evidence Innovative firms use more frequent less

flexible rule

Implications Innovative firms have specific requirements

Patent laws

Mechanisms Innovation protection motivates innovation Protection hinders further development

Empirical evidence 30% of innovations are patent protected Case studies Patent changes

Implications Dependent on technological advancement

Culture

Mechanisms Innovation attitude Cooperation attitude

Empirical evidence Different strengths in innovation processes Different institutions and laws Different interaction cultures

Implications Firm organisation and policy has to be

adapted

Industrial structure

Mechanisms Ideas and cooperation

Empirical evidence Case studies with path-dependence Spillovers

Implications Firm location Regional policy

Local Cluster Concepts, Facts and Theory The Topic Concepts Theory Facts Thomas Brenner DIMETIC October 2007.

Documents

local policy slide

local cluster concepts

causescharacterises

agglomeration of firms

lc cooperation id

success success

porter learning regions

different concepts