Presentation on theme: "International Technological Specialization in Important Innovations: Some Industry-Level Explorations Carolina Castaldi* and Bart Los** *University of."— Presentation transcript:
International Technological Specialization in Important Innovations: Some Industry-Level Explorations Carolina Castaldi* and Bart Los** *University of Utrecht & GGDC, **University of Groningen & GGDC EUKLEMS Consortium Meeting (Brussels, March 16 – 17, 2007) This project is funded by the European Commission, Research Directorate General as part of the 6th Framework Programme, Priority 8, "Policy Support and Anticipating Scientific and Technological Needs".
2 Introduction Lisbon agenda: goals with respect to dynamism and competitiveness of European economy. Innovation is a key factor Problem: innovation is hard to measure R&D expenditures are input indicators Surveys (CIS) sometimes subjectively filled out Patent counts imperfect measure Objectives of this project: Add to the literature on patent-related innovation measures Gain industry-specific knowledge about the ability of European countries to generate important innovations, relative to the U.S., Japan and Asian Tigers.
3 Measures of innovation output: patent indicators Body of literature on patents as output indicator (Schmookler, Scherer, Griliches, etc). Conclusion: patents useful but noisy indicator of innovation Patents very heterogeneous in importance (Hall, Pakes, Schankerman, Harhoff, etc.) In some industries, patenting is not seen as the most appropriate method to protect intellectual property (Cohen, Walsh, Nelson) Patent offices are not always functioning as they should, with imperfect examination procedures of prior art (Jaffe, Lerner) Citation counts can help in identifying important indicators (Trajtenberg, Jaffe, Hall)
4 Raw Patent Counts per Country Table 1 (p.h.w.: per 10,000 hrs worked) In 1998: HU: 10.6; CZ: 2.4; PL: 1.0
5 Problems to cope with… Point of departure: patents that receive more citations in subsequent patents are more important Problem 1: Patenting behavior varies across industries Problem 2: Citation behavior varies over time Problem 3: Citations are not received immediately Important innovations determined by constructing citation-based rankings by industry and year of grant for all patents issued; Distinction between important innovations and other innovations based on stylized fact concerning frequency distributions.
6 Stylized fact: Fat tails Curved part: lognormally distributed Linear part: Pareto distributed Hill estimator for fatness of tail: erratic behavior if observations not Pareto distributed Drees-Kaufmann procedure to estimate cut-off point Important innovations act as focal point for subsequent research (Silverberg & Verspagen, Sanditov) Bootstrapping to obtain confidence intervals
7 Data Sources NBER Patent-Citations Datafile Numbers of citations ( ) to all utility patents granted by USPTO in Our subset: (>2.4M patents, of which 1.0M to non-US inventors) Country of first inventor USPTOs PATSIC-CONAME Database Industry of manufacture (OTAF: 42 industries) Fractional counting in case of multiple OTAF codes Matching to 20 EUKLEMS industries 26 countries
8 Proportions of Important (Patented) Innovations by Industry (averages, )
9 Proportion of Important Innovations over Time (all manufacturing) (unweighted averages of industry-specific proportions)
10 Technology Life Cycles (number of important innovations: ) electronics machinery computers RTV aircraft ins wire ships metal prod oil food
11 Contributions to Worldwide Innovation (by period)…
12 …. and Contributions to Europes Important Innovations (by period)
13 Specialized in Important Innovations? US: 2.0 – 2.6%; Taiwan: 0.0 – 0.65%
14 Specialized in Important Innovations? (industry-level results, ) chemicals plastics cars aircraft
15 Further research Use of OECD PatStat database on international patent citations instead of NBER database More systematic analysis of distribution of cut-off point estimator Industry-of-use instead of industry-of-manufacture (Johnsons concordance), to link innovation indicator to EUKLEMS productivity indicators Study of relationship between important innovations and industry profitability using core EUKLEMS data Investigations to see whether techniques can be found to reduce time lag in identification process