THE PRODUCTIVITY OF SCIENCE: A CROSS COUNTRY ANALYSIS By G. Crespi and A. Geuna SPRU-University of Sussex Workshop on Measuring the Impact of Science INRS; Montreal June, 2004
Structure of the presentation The economic impact of public research; The productivity of science; First results of the cross country analysis: – Lag structure; – Spillovers; A first look to relative productivity changes; Conclusions.
The economic impact of scientific research Contribution of scientific research to industrial innovation: Jaffe, 1989; Mansfield, 1991; Narin, Hamilton, & Olivastro, 1997; Klevorick et al., 1995; Contribution of scientific research to productivity growth: Adams (1990); University-industry relationships; Technology Transfer and IPR.
The Productivity of science The productivity of science: – Adams & Griliches (1996;1998) Science funding and research outputs: 1. Time lag structure; 2. National returns and International spillovers; 3. TFP ~ catching up or falling behind.
A cross country analysis 1 Data: – 14 countries; – ; – HERD (problems!!); – Publications and citations (problems!!); => First results that underestimate the real return and limit the cross-country comparison due to data shortcomings.
A cross country analysis 2 1. Time lag (Polynomial Distributed Lag and Almond Model): – Length of the lag and then the degree of the polynomial function; – Publications: 6 years, 50% of impact by year 4; – Citations: 7 years, 50% of impact by year 5.
A cross country analysis 3 2.National returns and International spillovers: – HERD – Non-HERD %, – Time trend, – Country level fix effect, – International spillovers: ω= number of international co-authorships between countries i and l, divided by the total number of international co-authorships carried out by country l with the other countries in the sample.
A cross country analysis 3
Relative productivity analysis 1 TFP as measure of the overall organisational productivity of the science system; – Ranking (?); – Productivity growth rate: Convergence to the US frontier; Catching up with the UK.
Relative productivity analysis 1
Conclusions We were able to estimate a stable lag structure for publications (6Y) and citations (7y). We found constant return to scale at the international level, given data limitations this is a lower bound estimation of the returns. There is some evidence of a process of convergence in the productivity of science as measured by the outputs in the SCI-ISI.
Some of the limitations inherent in working at the cross-country level (such as the inability of working with field data) can be solved developing country comparisons on the basis of country and field level data. – The UK study