Towards a quantification of innovation systems Manuel Mira Godinho ISEG/UTLisbon Presentation to the Tampere 5 June 2008
Part 1 Lecture’s Topic Part 2 Conceptual Framework Part 3 Method Part 4 Cluster Analysis Part 5 Conclusions
Part 1 / Lecture’s Topic Is it possible to measure the development and maturity of a NIS? What specific technique can be used for that? Can we apply it to both the advanced, the emerging and the developing economies?
1st step Mapping NISs 2nd step With output of step 1: put forward a NIS taxonomy
Example of N.I.S. Mapping
Part 2 / Conceptual Framework
Different NIS Concepts Freeman (1987) organization of R&D in firms and role of government in Japan Nelson (1988) high tech sectors and R&D system Lundvall (1988) Inter-firm and user-producer interactions (...)
‘NIS’ emerged in the literature as a qualitative concept Is quantification possible / desirable? Possible: YES Desirable…YES, but! … caution needed in the analysis Each NIS Idiosyncratic
How to Quantify NISs? Concentrate on NIS “characteritics” Derivate quaintifiable “DIMENSIONS” (D1 to D8)
What a “NIS” is? “NIS” is a “system” “whole” more than “the parts” (Sources of increasing returns … ) - Knowledge spillovers - Network economies - Dynamic economies of scale - Agglomeration economies
NIS comprehends: -Actors (diversity, roles, behaviours, strategies) -Their interactions (linkages, channels, system density) -Institutions (with given functions, enable or limit innovation and diffusion)
NIS purposes -Allocation of resources for innovation and diffusion -Speed up accumulation and distribution of knowledge -Provide a favourable regulatory framework
NIS peformances a) learning, accumulation of capabilities … b) … innovation, diffusion … c) …. growth, development, sustainability
‘Innovation’ vs. ‘Diffusion’ in N.I.S. trade-off or complementarity ? In some NIS ‘diffusion’ more important than ‘innovation’ (in the limit ‘innovation’ = 0, but even in this case we can speak of ‘NIS’)
Part 3 - Method Decide what are the relevant n Dimensions Decide what variables shall be used for each D All indicators standardized Aggregate 2-5 indicators into each relevant D Map D1 to D8 into bi-dimensional space 8 Dimensions object of cluster analysis
8 NIS dimensions defined –Market opportunities –Institutional conditions –[intangible and tangible] Accumulation –S&T Opportunities –Economic structure –External communication –Diffusion –Innovation In order to materialise such 8 NIS dimensions 27 individual indicators selected
Dimension 1 - Market Opportunities - Income per capita - Overall GDP size - Population density Dimension 2 - Institutional conditions - GINI index (1/G) - Youth of population - Life expectancy - Corruption index Dimension 3 – ( Intangible and tangible) Accumulation - Education expenditures / GDP - Education / Population - GERD / GDP - GERD /Population - GF Investment rate / GDP Dimension 4 – S&T Opportunities - Researchers / Population - Scientific Papers/ Population - First University Degrees in S&E / Population
Dimension 5 - Economic structure - Value Added in High-Tech & Medium High-Tech Activities (% of MVA) - High-Tech & Medium High-Tech Exports (%) - Sales of home-based top 750 global R&D companies / GDP Dimension 6 - External communication - (Exports + Imports) / GDP - (Inward + Outward stocks of FDI) / GDP - Bandwidth in international connections (bits per Capita) Dimension 7 - Diffusion - Personal Computers / Population - Internet Users/ Population - Cellular Phones/ Population - ISO ISO Certificates/ Population Dimension 8 - Innovation - US Patents/ Population - EU Trademarks / Population
69 Countries in the analysis Developed + emerging + developing economies OECD economies “Asian tigers” included All countries > 20 M inhabitants Sample: > 87% of the world population
2 time moments
Part 4 - Cluster analysis The object of the analysis was a matrix with 69 countries in the sample as the individual ‘cases’ 8 NIS dimensions as the ‘variables’ to be analysed Cluster analysis apllied to 2000/1 and 2005/6 9 different clustering algorithms Results compared for stability
2000/1 2 Megaclusters 4 main Clusters
C1 C2 C4 C3
4 Main Clusters 2000
<C1 Sudan Ethiopia D.R. Congo Kenya Nigeria Myanmar Pakistan Indonesia Peru Colombia Bulgaria Egypt Turkey India Romania Ukraine Venezuela Argentina Russia Iran (I.R.) Tanzania Bangladesh Algeria Morocco Viet Nam Philippines Ukraine Latvia Lithuania Poland Estonia Slovak Republic Slovenia Hungary Czech Republic South Africa Thailand Chile Brazil China Latvia Mexico Greece Cyprus Malaysia Portugal Malta C2A C2>
C4 C3
2000: Cluster 2 and Subgroup 2A Cluster 2 Group 2A
2005/6 Again 2 Megaclusters 4 main Clusters
2005/6 But… Catching Up is Visible. 5 of the 8 in previous group 2A move now to Cluster 3. In 3 (out of 9) clustering simulations the number of countries moving to C3 is 5+11
C3 C4 C1 C2 (n=21) 6/9
C3 C4 C2 (n=11) C1 3/9
X Y
4 Main Clusters 2006
C2 C2Y C2X China Mexico Russia Thailand Turkey Ukraine South Africa Brazil India Argentina Malta Portugal Malaysia Poland Lithuania Latvia Cyprus Greece Bulgaria Chile Romania
Overall Comparison of NIS Evolution Possibility of establishing a ranking Rank measure ≡ NIS map area
Part 5 > Conclusions
Methodological aspects Quantification possible, but... Need of appropriate indicators e.g. on networking, on innovation in low and medium tech sectors, even detailed R&D data lacking
Further conclusions: policy application Responds to policy demand for guidance Comparability/benchmarking Summary measures Scoreboards have been produced But criticized: loss of information, simplification