Findings from the Cluster Meta-Study Dr. Claas van der Linde Research Institute for International Management, University of St. Gallen October 4, 2002 This presentation provides an outline of a presentation and is incomplete without the accompanying oral commentary and discussion. It draws on research in progress conducted by Dr. Claas van der Linde and Prof. Michael E. Porter, both Harvard Business School. 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 Claas van der Linde or Michael E. Porter. Additional information may be found at the website of the Institute for Strategy and Competitiveness, www.isc.hbs.edu
Cluster Meta-Study The Need for New Methodologies A wealth of individual case studies and cluster reports often more qualitative than quantitative not standardized in methodology or depth of analysis Little systematic, quantitative evidence on the population of clusters as a whole usually focused on a country or a specific region often a narrow focus on some thematic aspect suffer from incomplete data sources and methods The Cluster Meta-Study A multiyear effort to aggregate data from existing cluster publications to learn more about the locational, economic, and competitive characteristics of clusters, the reasons behind their competitiveness, and their patterns of evolution over time 2
Meta-Analysis versus Meta-Study Meta-Analysis Meta-Study Based on experimental studies Primarily quantitative data Underlying studies are quite uniform Focus on reporting studies Tests the impact of one or several variables on a dependent variable Seeks to prove a causal relationship Based on non-experimental information Codifies primarily qualitative data Underlying studies are not uniform Clusters as the unit of analysis Captures the universe of available studies Descriptive statistics and associations rather than causal relationships Given the medical connotations of meta-analysis we employ the term metastudy for our purposes 3
The Cluster Template Cluster Description Biographical Biographical information information Cluster Cluster characteristics characteristics Location Geographic Geographic boundaries boundaries Regional Regional economic economic development development State State of of Competitiveness Indicators Indicators of of competitiveness competitiveness Determinants of of Competitiveness Factor Factor conditions conditions Demand Demand conditions conditions Related Related & supporting supporting industries industries Context Context for for strategy strategy and and rivalry rivalry Evolution Path Path Cluster Cluster birth birth Cluster Cluster decline decline A tool for systematically profiling clusters and assessing their competitiveness to allow comparison across disparate studies 4
The Cluster Meta-Study Clusters in the Database 25 Developed Nations 24 Developing Nations 2 Andorra 72 Italy 3 Brazil 5 Morocco 4 Australia 4 Japan 1 Chile 1 Nepal 7 Austria 6 Netherlands 1 Columbia 1 Pakistan 1 Belgium 10 New Zealand 2 Costa Rica 9 Palestinian Auth. 1 Bermuda 1 Norway 3 Ecuador 6 Philippines 13 Canada 4 Portugal 1 Estonia 1 Poland 34 Denmark 3 Singapore 1 Ghana 2 South Africa 11 Finland 6 Spain 106 India 1 Taiwan 96 France 5 Sweden 4 Jordan 1 Thailand 31 Germany 13 Switzerland 4 Kenya 1 Turkey 11 Hong Kong 168 United Kingdom 3 Lebanon 2 Venezuela 2 Ireland 153 United States 2 Malaysia 6 Israel 8 Mexico 664 Clusters from Developed Nations 169 Clusters from Developing Nations Information on 833 clusters from 49 countries, including 24 developing nations Based on 375 published and unpublished cluster-oriented books, reports, and articles A total of 26,035 observations of which 14,069 were evaluated quantitatively 5
Sample Composition Clusters in the Database Materials/Metals (6.6%) Primary Goods (66.5%) Forest Products (2.8%) Primary Services (9.1%) Machinery (3.9%) Petroleum/Chemicals (5.4%) Inputs (2.6%) Semiconductors/Computers (4.8%) Support Services (1.3%) Multiple Business (12.7%) Multiple Stages (16.7%) Transportation (7.5%) Power Gen. & Distribution (2.3%) Office (0.6%) Telecommunications (2.5%) Defense (0.6%) Food/Beverage (8.8%) Textiles/Apparel (14.6%) Housing/Household (11.6%) Health Care (5.8%) Personal (3.1%) Entertainment/Leisure (10.1%) 0% 1% 2% 3% 4% 5% 6% 7% 8% 9% 10% 11% 12% 13% 14% 15% 16% Source: Cluster Meta-Study, (www.isc.hbs.edu) 6
Cluster Size Employment in Cluster or Core Industry 4 38.1% 35.0% Percent of Total 3 25.0% 2 15.0% 1 5.0% 12.3% 8.8% 7.0% 5.0% 2.0% 26.9% less than 5,000 jobs (174 obs.) 5,000 to 10,000 jobs (56 obs.) 10,000 to 15,000 jobs (40 obs.) 15,000 to 20,000 jobs (32 obs.) 20,000 to 25,000 jobs (23 obs.) 25,000 to 30,000 jobs (9 obs.) more than 30,000 jobs (123 obs.) Source: Cluster Meta-Study, (www.isc.hbs.edu) 7
Cluster Size Median Employment by Country 30,000 26,000 25,000 23,831 21,000 Employment 20,000 15,000 10,000 5,000 5,576 5,000 5,000 2,300 5,000 0 Canada (11 obs.) U. S. (72 obs.) U.K. (103 obs.) Italy (60 obs.) Denmark (17 obs.) Germany (17 obs.) France (75 obs.) India (58 obs.) Source: Cluster Meta-Study, (www.isc.hbs.edu) 8
Cluster Size Number of Companies in Cluster or Core Industry 45.0% 4 41.6% 35.0% Percent of Total 3 25.0% 2 15.0% 1 5.0% 13.9% 9.4% 2.6% 3.7% 3.4% 25.4% less than 100 firms (159 obs.) 100 to 200 firms (53 obs.) 200 to 300 firms (36 obs.) 300 to 400 firms (10 obs.) 400 to 500 firms (14 obs.) 500 to 600 firms (13 obs.) more than 600 firms (97 obs.) Source: Cluster Meta-Study, (www.isc.hbs.edu) 9
Cluster Boundaries 45.0% 41.8% 4 35.0% Percent of Total 3 25.0% 2 15.0% 18.9% 19.9% 1 6.1% 6.8% 6.1% 5.0% 0.4% within a city within a within part of a within a state across state within a nation across national (295 obs.) metropolitan state (43 obs.) borders (43 obs.) borders area (140 obs.) (48 obs.) (3 obs.) (133 obs.) Source: Cluster Meta-Study, (www.isc.hbs.edu) 10
State of Cluster Development Compared to Competing Clusters 45.0% 4 35.0% Established 39.0% Emerging or Unravelling Percent of Total 3 25.0% 2 15.0% 1 14.7% 20.5% 9.0% 8.0% 8.8% 5.0% among world's three most highly developed (85 obs.) highly developed (225 obs.) critical mass present (118 obs.) partly developed (52 obs.) lacking critical mass (46 obs.) rudimentary (51 obs.) Source: Cluster Meta-Study, (www.isc.hbs.edu) 11
Cluster Competitiveness Compared to Competing Clusters 3 27.3% 25.0% Percent of Total 2 15.0% 1 11.0% 9.3% 11.4% 18.2% 18.4% 5.0% 4.4% considered among world's 3 among world's 10 internationally nationally weak uncompetitive world's strongest most competitive most competitive significant significant (79 obs.) (19 obs.) (47 obs.) (40 obs.) (49 obs.) (78 obs.) (117 obs.) Source: Cluster Meta-Study, (www.isc.hbs.edu) 12
Cluster Competitiveness Change in Competitive Position 3 25.0% 25.0% 25.0% 20.8% Percent of Total 2 15.0% 1 7.8% 12.5% 6.8% 5.0% 2.1% rapidly gaining position (15 obs.) modestly gaining position (24 obs.) slowly gaining position (48 obs.) holding position (48 obs.) slowly losing position (40 obs.) modestly losing position (13 obs.) rapidly losing position (4 obs.) Source: Cluster Meta-Study, (www.isc.hbs.edu) 13
Cluster Age 4 37.6% 35.0% 3 Percent of Total 25.0% 2 15.0% 1 5.0% 14.6% 4.9% 9.8% 11.7% 21.5% before 1750 (30 obs.) between 1750 and 1800 (10 obs.) between 1800 and 1850 (20 obs.) between 1850 and 1900 (24 obs.) between 1900 and 1950 (44 obs.) between 1950 and 2000 (77 obs.) Source: Cluster Meta-Study, (www.isc.hbs.edu) 14
Cluster Age Median Year of Cluster Establishment by Country 2000 1950 1949 1950 1955 1900 Year (Median) 1850 1800 1780 1800 1804 1820 1823 1750 1700 1650 U. K. France Germany Switzerland Italy USA India New Zealand (5 obs.) (15 obs.) (24 obs.) (13 obs.) (12 obs.) (44 obs.) (56 obs.) (5 obs.) Source: Cluster Meta-Study, (www.isc.hbs.edu) 15
Competitiveness and Cluster Age 4 3 Current Competitiveness 2 1 0-1 -2-3 y = -1E-05x + 0.8535 R 2 = 2E-06-4 1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000 Year of Cluster Establishment No relationship between a cluster's age and its competitiveness! Source: Cluster Meta-Study, (www.isc.hbs.edu) 16
Determinants of Cluster Competitiveness Context for Strategy and Rivalry Other (Chance, Government) Factor (Input) Conditions Open and vigorous competition among locally based rivals A local context that encourages investment and sustained upgrading Demand Conditions High quality, specialized inputs available to firms: human resources capital resources physical infrastructure administrative infrastructure information infrastructure scientific and technological infrastructure natural resources Related and Supporting Industries Availability of capable, locally based suppliers and firms in related fields A core of sophisticated and demanding local customer(s) Unusual local demand in specialized segments that can be served nationally and globally Customer needs that anticipate those elsewhere 17
Dominant Basis of Current Cluster Competitiveness Most Important Determinant (Percent of Total) 5 45.0% 4 35.0% 3 25.0% 2 15.0% 1 5.0% 43.0% Factor Conditions (68 obs.) 24.7% Demand Conditions (39 obs.) 13.3% 13.3% Related & Supporting Industries (21 obs.) Context for Strategy and Rivalry (21 obs.) 5.7% Other (Chance, Gov't) (9 obs.) Source: Cluster Meta-Study, (www.isc.hbs.edu) 18
Dominant Basis of Current Cluster Competitiveness Competitive versus Uncompetitive Clusters Most Important Determinant (Percent of Total) 6 5 4 3 2 1 38.3% 21.0% 16.0% 21.0% 3.7% Cluster among world's 10 most competitive clusters (81 obs.) 52.2% 30.4% 8.7% Factor Conditions Demand Conditions Related & Sup. Industries 8.7% Cluster weak or uncompetitive (23 obs.) Context for Strategy and Rivalry Other (Chance, Government) Source: Cluster Meta-Study, (www.isc.hbs.edu) 19
Dominant Basis of Current Cluster Competitiveness Competitive versus Uncompetitive Clusters Competitive Clusters Uncompetitive Clusters Strategy & Rivalry (21.0%) Other (Chance, Government) (3.7%) Other (Chance, Government) (8.7%) Factor Conditions (38.3%) Demand Conditions (21.0) Factor Conditions (52.2%) Demand Demand Conditions Conditions (30.4%) (30.4%) Related Related and and Supporting Supporting Industries Industries (16.0%) (16.0%) Related and Supporting Related and Industries Supporting (8.7%) Industries (8.7%) Competitive clusters may rely on any part of the full diamond, uncompetitive clusters usually only on factor conditions! Source: Cluster Meta-Study, (www.isc.hbs.edu) 20
Dominant Basis of Current Cluster Competitiveness 3.0 Competitiveness versus Dominant Determinant Mean Competitiveness 2.0 1.0 0.0-1.0-2.0 1.1 0.4 0.4 0.3 1.7 Sample Mean -3.0 Factor Conditions Demand Conditions Related & Supporting Industries Context for Strategy and Rivalry Other (Chance, Gov't) Clusters which relied primarily on rivalry are significantly more competitive than clusters which relied predominantly on factor conditions! Source: Cluster Meta-Study, (www.isc.hbs.edu) 21
Dominant Basis of Current Cluster Competitiveness 18 Summary: Frequency versus Associated Competitiveness 1.7 16 14 12 10 8 6 4 2 43.0% Factor Conditions Demand Conditions Related & Sup. Industries Context for Strategy and Rivalry Other (Chance, Government) 24.7% 13.3% 13.3% 5.7% Most Important Determinant (Frequency of Occurrences) 0.4 0.4 1.1 0.3 Competitiveness Associated with Determinant (Mean Competitiveness on a Scale from -3 to +3) Clusters rely most often on factor conditions, but those which rely primarily on rivalry are most competitive! Source: Cluster Meta-Study, (www.isc.hbs.edu) 22
Dominant Basis of Current Cluster Competitiveness Developed versus Developing Nations 7 66.7% Most Important Determinant (Percent of Total) 6 5 4 3 2 1 35.3% 26.1% 15.1% 16.0% 7.6% Clusters based in Developed Nation (119 obs.) 20.5% 7.7% Factor Conditions Demand Conditions Related & Sup. Industries Context for Strategy and Rivalry Other (Chance, Government) 5.1% Clusters based in Developing Nation (39 obs.) Source: Cluster Meta-Study, (www.isc.hbs.edu) 23
Dominant Basis of Current Cluster Competitiveness Relative to Cluster Age Most Important Determinant (Percent of Total) 45.0% 4 35.0% 3 25.0% 2 15.0% 1 5.0% 31.0% 20.7% 6.9% 31.0% 10.3% Cluster established before 1900 (29 obs.) 39.0% 22.0% 18.6% Factor Conditions Demand Conditions 16.9% Cluster established after 1900 (59 obs.) Related & Sup. Industries Context for Strategy and Rivalry Other (Chance, Government) 3.4% Source: Cluster Meta-Study, (www.isc.hbs.edu) 24
Most Important Reasons for Cluster Establishment 45.0% 4 39.8% 35.0% Percent of Total 3 25.0% 2 15.0% 18.8% 15.1% 26.3% 1 5.0% Factor Conditions (74 obs.) Demand Conditions (35 obs.) Related & Supporting Industries (28 obs.) Context for Strategy and Rivalry (0 obs.) Other (Chance, Gov't) (49 obs.) Source: Cluster Meta-Study, (www.isc.hbs.edu) 25
Most Important Reasons for Cluster Establishment Relative to Cluster Age 45.0% 4 38.6% 37.7% FC Percent of Total 35.0% 3 25.0% 2 15.0% 22.8% 12.3% 26.3% 29.0% 14.5% 18.8% DC RSI CSR Other 1 5.0% Cluster established before 1900 (57 obs.) Cluster established 1900 or later (69 obs.) Source: Cluster Meta-Study, (www.isc.hbs.edu) 26