GDP / Capita purchasing power standards of EU average

(1)

Cluster Mapping in

Europe and the United States

Christian H. M. Ketels, PhD Institute for Strategy and Competitiveness

Harvard Business School Trend Chart Workshop Brussels, Belgium 15 November 2005

This presentation has benefited from Professor Michael E. Porter’s articles and books and ongoing research at the Institute for Strategy and Competitiveness as well as joint work with Professor Örjan Sölvell at the Center for Strategy and Competitiveness, Stockholm School of Economics. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means - electronic, mechanical, photocopying, recording, or otherwise - without the permission of the author

Additional information on the Institute for Strategy and Competitiveness is available at www.isc.hbs.edu Additional information on the Center for Strategy and Competitiveness is available at www.sse.edu/csc

(2)

The Role of Cluster Mapping

• Provide a precise language for discussing clusters and their role in regional economies

• Provide data for regional economies to develop competitiveness strategies reflecting their individual cluster portfolios

• Enable regional clusters to systematically compare their size and profile over time and with peers in other locations

• Guide the use of policy instruments tied to the presence of clusters across locations

• Cluster mapping is a key element in moving cluster-based economic

policy to the next level

(3)

Cluster Policy Approaches

• Cluster effort often based on national programs

• Strong role of government in initiating cluster efforts

• Lower level of specialization across regional economies

• Business environments tend to be strong on factor input

conditions, often weaker on context for strategy and rivalry

• Cluster effort often based on national programs

• Strong role of government in initiating cluster efforts

• Lower level of specialization across regional economies

• Business environments tend to be strong on factor input

conditions, often weaker on context for strategy and rivalry

• Cluster efforts based on regional initiatives

• Strong role of private sector from the outset of cluster efforts

• Many regional economies highly specialized around strong clusters

• Business environments tend to be very open to cross-regional

competition and have access to strong factor input conditions

• Cluster efforts based on regional initiatives

• Strong role of private sector from the outset of cluster efforts

• Many regional economies highly specialized around strong clusters

• Business environments tend to be very open to cross-regional

competition and have access to strong factor input conditions Europe

Europe United States United States

(4)

Use of Cluster Mapping

United States

• Identification of regional clusters

• Assessment of economic performance of regional clusters

• Development of regional strategies to mobilize clusters

• Analysis of the relationship between cluster presence and regional economic performance

Europe

• Intentions as above

• Intention to use cluster definitions to guide public policy programs

(5)

Cluster Mapping Approaches

HBS Cluster Mapping Project

• Based on actual co-location of industries; revealed impact of sum of locational factors on company decisions

• Use of U.S. data because the U.S. economy has been exposed to free cross- regional competition among the locations for the longest time

– “a peek into the future of other locations”

Alternatives/complements

• Input – output relationships; supplier relationships

• Cross-company/institution career paths; social networks

• Co-publication/citation data; knowledge spill-overs

(6)

HBS Cluster Mapping Project

• Use of employment data at the 4-digit industry level for regional economies

• Calculation of regional concentration per industry across the U.S.

– No concentration: Local industries

– Significant concentration: Traded clusters and natural-resource driven clusters

• Calculation of correlation patterns among industries in the traded clusters-category

• Based on correlation patterns identification of 41 cluster groups (and

>200 sub-cluster groups) that industries get assigned to

– Narrow cluster definition: Each industry allocated to one cluster

– Broad cluster definition: Industries can be allocated to more than one

cluster

(7)

Traded Clusters

Traded Clusters Local Clusters Local Clusters Local Clusters

Natural

Resource-Driven Industries

Natural Natural Resource

Resource- -Driven Driven Industries Industries

30.5%

0.9%

$45,511 129.7%

4.3%

144.1 21.3 590 30.5%

0.9%

$45,511 129.7%

4.3%

144.1 21.3 590

68.8%

2.4%

$29,010 82.7 3.6%

79.3 1.3 241 68.8%

68.8%

2.4% 2.4%

$29,010

$29,010 82.7 82.7 3.6% 3.6%

79.3 79.3

1.3 1.3 241 241

0.7%

-1.2%

$33,066 94.3 1.8%

140.1 7.0 48 0.7% 0.7%

-1.2% - 1.2%

$33,066

$33,066 94.3 94.3 1.8% 1.8%

140.1 140.1

7.0 7.0 48 48

Share of Employment Employment Growth Rate,

1990 to 2002 Average Wage Relative Wage Wage Growth

Relative Productivity

Patents per 10,000 Employees Number of SIC Industries

Note: 2002 data, except relative productivity which uses 1997 data.

Source: Prof. Michael E. Porter, Cluster Mapping Project, Institute for Strategy and Competitiveness, Harvard Business School

Composition of Regional Economies

United States, 2002

(8)

Plastics

Oil and Chemical Gas

Products

Pharma- ceuticals

Power Generation Aerospace Vehicles &

Defense

Lightning &

Electrical Equipment Financial

Services

Publishing and Printing

Entertainment

Hospitality and Tourism

Transportation and Logistics

Information Technology

Communi- cations Equipment

Medical Devices

Analytical Instruments Education

and Knowledge

Creation Apparel

Leather and Sporting

Goods

Agricultural Products

Processed Food

Furniture Building

Fixtures, Equipment

and Services

Cluster Overlap in the United States Economy

Common Industries Across Broad Traded Clusters

Note: Clusters with borders or identical colors/shading except gray have at least 20% overlap of industries by number in both directions

Sporting, Recreation and Children’s

Goods

Business Services

Distribution Services Fishing &

Fishing Products

Footwear

Forest Products

Heavy Construction

Services

Jewelry &

Precious Metals

Construction Materials

Prefabricated Enclosures Textiles

Tobacco

Heavy Machinery Aerospace

Engines

Automotive

Production Technology

Motor Driven Products Metal

Manufacturing

(9)

0 1 2 3 4

-50 0 50 100

Specialization of Regional Economies

Atlanta Metro Area

Percentage Share of

National Cluster Employment

in 2000

Percentage Change, 1990–2000

= 0–19,999 = 20,000–49,999 = 50,000–99,999 = 100,000+

Power Generation (1.8, 320.1)

Oil and Gas Agricultural Products Leather Products

Heavy Construction Services

Heavy Machinery Processed Food

Analytical Instruments

Production Technology Metal

Manufacturing

Education and Knowledge Creation Distribution

Services Financial

Services

Transportation and Logistics (4.1, 74.7)

Business Services

Jewelry and Precious Metals Prefabricated

Enclosures

Furniture Lighting and

Electrical Equipment

Apparel

Hospitality and Tourism

Pharmaceuticals and Biotechnology

Atlanta’s Average Share = 1.9%

Note: Uses narrow cluster definitions to avoid overlap

Source: Cluster Mapping Project, Institute for Strategy and Competitiveness, Harvard Business School

Motor Driven Products Aerospace Vehicles

and Defense

Aerospace Engines (0.5, 601.7)

Textiles Building Fixtures,

Equipment and Services

Sporting Products Automotive

IT Communications

Equipment

(10)

$15,000

$25,000

$35,000

$45,000

$55,000

50 100 150 200 250 300

Average Regional Wage, 2001

Share of Traded Employment in Strong Clusters (LQ > .8), Broad Cluster, 2001

y = 96.736x + 16218 R

²

= 0.377 New York, NY

Bay Area, CA

Boston, MA

Determinants of Regional Prosperity

Cluster Strength and Wage Levels, U.S. Regions

Source: County Business Patterns; Michael E. Porter, The Economic Performance of Regions”, Regional Studies, Vol. 37, 2003

(11)

Specialization and GPP / Capita

y = 336,5x + 35,336 R

²

= 0,4358 0

20 40 60 80 100 120 140 160 180

0% 5% 10% 15% 20%

Percent of employment in specialized clusters (SQ>2) G D P / C a p ita p u rc h a s in g pow e r st andar d s of E U av er age

Praha City

Bratislava

Determinants of Regional Prosperity

Cluster Strength and GDP per Capita, EU-10 Regions

Source: Solvell/Ketels/Frederiksson, Regional clusters in the EU-10, 2005

(12)

0.50 0.75 1.00 1.25

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

Determinants of Regional Prosperity

Traded Cluster Specialization and Relative Wage Levels: Ohio

Relative Cluster Wage

Relative Employment (LQ) by Traded Cluster Financial Services

Note: Uses narrow cluster definitions to avoid overlap; bubble size proportional to employment bracket Source: Cluster Mapping Project, Institute for Strategy and Competitiveness, Harvard Business School

Automotive Metal Manufacturing

4.29% of U.S. Employment

U.S. average cluster wage Business Services

Production Technology

y = 0.1902Ln(x) + 0.9874

R

²

= 0.3403

(13)

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Determinants of Regional Prosperity

Level versus Mix Effect, U.S. Regions

Cluster Wage Level Effect as % of Wage Gap, 2001

U.S. Economic Areas

Median: 74.2%

(14)

Determinants of Regional Prosperity

Change in Cluster Specialization and Wage Growth, U.S. States

2.0%

2.5%

3.0%

3.5%

4.0%

4.5%

5.0%

5.5%

-0.06 -0.04 -0.02 0.00 0.02 0.04 0.06 0.08 0.10

Annual Regional Wage Growth Rate, 1990-2001

Change of Cluster Employment GINI, 1990-2001

y = 8.7905x + 3.6107

R²= 0.2626 P-value = .0001

MA NY

Economy becoming less specialized

Economy becoming more specialized

AK

CA CT CO

DE DC

FL GA

HI ID

IL

IN IA

KS

KY

LA

MN

MO NV MT

NJ NC

ND

OK OR

PA RI

SC

TX WA VA

WV WI

WY AL

AR

(15)

Explaining Average Regional Wages

Multiple Regression Model

Independent Variable Independent Variable

• Total regional employment

• Patents per capita

• Patentor concentration

• Share of strong clusters in regional employment

• Cluster breadth

• Total regional employment

• Patents per capita

• Patentor concentration

• Share of strong clusters in regional employment

• Cluster breadth

Effect Effect

Positive, significant Positive, significant Negative, significant Positive, significant

Positive, significant Positive, significant Positive, significant Negative, significant Positive, significant

Positive, significant

Dependent variable: Regional Average Wage

Note: Regression uses 2001 data for 172 U.S. economic areas

Source: Michael E. Porter, The Economic Performance of Regions”, Regional Studies, Vol. 37, 2003

Explained Variation (adjusted R

²

): 72.8%

(16)

Comparative Data on European Clusters

Stockholm Cluster Portfolio

0%

10%

20%

30%

40%

50%

60%

-15% -10% -5% 0% 5%

Change of Share in National Cluster Employment, 1995-2003

Stockholm Share of National Cluster Employment, 2003: 22.9%

Change in Stockholm’s overall share of National Cluster Employment: -0.5%

Note: Bubble size is proportional to employment levels Source: Statistics Sweden (2005), author’s calculations

Biopharmaceuticals

Financial Services

Business Services Communication Equipment

Information Technology

Distribution Services

Education & Knowledge Creation

Heavy Construction Services Tourism Publishing & Printing

Analytical Instruments

Transportation & Logistics Share in National

Cluster Employment,

2003

(17)

Region

Region Cluster Cluster Employment Employment

Schleswig-Holstein (DE) Västsverige (SE)

Hamburg (DE) Etelä-Suomi (SF) Stockholm (SE)

Östra Mellansverige (SE)

Mecklenburg-Vorpommern (DE) Warminsko-Mazurskie (PL) Norra Mellansverige (SE) Oslo og Akershus (NO) Småland med öarna (SE) Warminsko-Mazurskie (PL) Norra Mellansverige (SE) Islands (IS)

Agder og Rogaland (NO) Länsi-Suomi (SF)

Schleswig-Holstein (DE) Västsverige (SE)

Hamburg (DE) Etelä-Suomi (SF) Stockholm (SE)

Östra Mellansverige (SE)

Mecklenburg-Vorpommern (DE) Warminsko-Mazurskie (PL) Norra Mellansverige (SE) Oslo og Akershus (NO) Småland med öarna (SE) Warminsko-Mazurskie (PL) Norra Mellansverige (SE) Islands (IS)

Agder og Rogaland (NO) Länsi-Suomi (SF)

Financial Services Automotive

Financial Services Forest Products Business Services Metal Manufacturing Hospitality and Tourism Processed Food

Metal Manufacturing Business Services Metal Manufacturing

Building Fixtures, Equipment and Services Forest Products

Fishing and Fishing Products

Oil and Gas Products and Services Metal Manufacturing

Financial Services Automotive

Financial Services Forest Products Business Services Metal Manufacturing Hospitality and Tourism Processed Food

Metal Manufacturing Business Services Metal Manufacturing

Building Fixtures, Equipment and Services Forest Products

Fishing and Fishing Products

Oil and Gas Products and Services Metal Manufacturing

60,423 43,168 42,420 40,722 38,283 28,706 26,538 21,831 21,240 17,966 16,995 14,431 13,674 11,931 10,752 10,090 60,423 43,168 42,420 40,722 38,283 28,706 26,538 21,831 21,240 17,966 16,995 14,431 13,674 11,931 10,752 10,090

Note: “3 Star” defined as >10.000 employees, > 10% of regional employment, and SQ > 2. Data set does not include Denmark and Russia Source: Institute for Strategy and Competitiveness, author’s calculations

3 STAR-Clusters in the Baltic Sea Region

(18)

Estonia

Latvia

Cyprus

Malta Praha Region; CZ

Praha City; CZ

Gdansk; PL Szczecin; PL

Warszawa; PL

Slovenia

Budapest; HU Miskolc; HU

Comparative Data on European Clusters

Transportation and Logistics Clusters in the EU-10

Source: Solvell/Ketels/Frederiksson, Regional clusters in the EU-10, 2005

(19)

Conclusion

• Cluster mapping is a tool, not a solution; it is critical for a more fact- driven discussion about cluster-based economic policy

• Cluster mapping efforts should be part of a wider cluster data infrastructure

– Cluster-specific business environment assessments – Impact assessment for cluster-based policy initiatives

• Creating this data infrastructure is a useful task for the European Commission; running cluster-based efforts themselves is not

• The available cluster mapping data suggests that Europe is in the

midst of a relocation process that has already proceeded much

further in the U.S.

(20)

Back-Up

(21)

Presence of Clusters Across Countries

Selected Countries

“Clusters are common and

deep”

“Clusters are limited and

shallow”

Note: EU members in red (EU-15) and blue (NMS), other countries in green; arrows indicate significant changes since 2002 Source: Global Competitiveness Report 2004-2005, World Economic Forum

Survey Question: “How Common Are Clusters In Your Country?”

Average of all 93 countries

1 2 3 4 5 6 7

Japan Finland

United States Italy Denm

ark Sweden

Ireland

United Kingd om

Canada Germany

Switzerland Norwa

y Austria

France Netherlan

ds Belgium

Portuga l

Australia Spain New Zealand

Lithuania Czech Repub

lic Sloven

ia Poland

Slovak Re pub

lic Latvia

Greece Estonia

Hunga ry

Malta