The Impact of R&D on Innovation and Productivity: Statistical Analyses & Case Studies in Intellectual Property

1. The Impact of R&D on Innovation and Productivity Professor Derek Bosworth Intellectual Property Research Institute of Australia Melbourne University

2. Methodologies • Case studies of the impact of R&D (i.e. studies of agricultural research laboratories) • Surveys of the effects of R&D and innovation (i.e. Community Innovations Survey) • Statistical analyses of large scale, micro (company level) data sets (i.e. Griliches’ and Hall’s work using COMPUSTAT)

3. Statistical method I • Market valuation function • MV = f (tangible assets,intangible assets, …) • MV = f (tangible assets,past R&D, …) • Answers question: • what value does the market place on the intangible assets of the company? • what value does the market place on the activities which generate intangible assets?

4. Statistical method II • Production function • output = f (capital,labour,past R&D, …) • TFP = f (past R&D, …) • Answers question: what role does R&D play in driving total factor productivity? • Methods I and II are consistent and give similar empirical results

5. Data used in statistical studies • Panel data • large number of companies • long periods of time • Explanatory variables • tangible inputs (capital stock, labour) • past R&D or the R&D stock • patents or the patent stock • pool of R&D

6. Overview: effects of R&D • Past R&D has a positive and large impact on productivity • R&D assets are valued by financial markets • The effects of R&D change systematically over time

7. Overview: role played by IP • Patents play a role (above and beyond R&D) in explaining firm performance, although the correlation is much weaker • Citation-weighted patents are slightly more informative than patents • Trade marks and other forms of IP are valued by the market

8. Overview: other key results • Competitor R&D reduces the returns to the company’s own R&D[?] • Company’s own R&D improves the ability of the company to innovate using inventions produced by others • The R&D pool is at least as important as the company’s own R&D

9. Average gross rate of return to R&D • Firm-level estimates • 27% - USA and Japan • (average, 24 studies) • 34% - Belgium, Canada, France and Germany • (average, 6 studies) • Industry-level estimates • 26% - USA and Japan • (average, 24 studies) • 34% - Belgium, Canada, France, Germany and UK • (average, 6 studies)

10. Risk and uncertainty • Results from large panel data sets are broadly consistent and robust • However, they tend to be much less consistent and robust in smaller samples • for particular sectors • for particular firms • Reason is that risk and uncertainty become important at the individual firm level

11. Sources and nature of risk • Risk comes from many sources • Each of the following has a probability distribution • R&D will find anything • the patent office will grant patent • a new or revised product will result • market accepts/adopts the new product • a higher quality competitor product is launched

12. Skewed nature of value • Each probability distribution may be quite simple (say normally distributed) • Probabilities at each stage interact - a highly skewed overall distribution of values of inventions • Empirical results demonstrate most inventions are worth very little and a few are worth very large amounts

13. Normal distribution % of population Value

14. Skewed distribution of values % of population Value

15. Inequality in the Value of Patented Inventions Cumulative value Cumulative number of patents: ranked from least to most valuable