Which Scientometric Indicators Best Explain National Performance of High-Tech Outputs? R. D. SheltonTarek Fadel WTEC ITRI Presented at the Collnet Conference, Ilmenau, Germany, Sept. 2014
Outline Input/Output Models of National Innovation Systems Trends in High Tech Outputs: Exports, Overall Manufacturing Strongest Correlations with Inputs: R&D Investments, Researchers, and with Intermediate Indicators: Papers, Patents But Which is the Cause and Effect?
Sales are the main way that investments can be recovered. There are also feedback loops, e.g. one going back from sales to BERD. Linear, simplified model of a national innovation system with indicators. Part of domestic manufacturing sales is exports.
The effect of U.S. and Japanese off-shoring of assembly to China is clear. Germany did much less of this. But, this double counts imported parts. World share of exports of high-technology products, on a cash basis, showing the dramatic rise of China to lead the world in 2005
China had not yet taken the world lead in Later we will extrapolate these trends. World share of manufacturing of high-technology products, on a value- added basis, from a new OECD-WTO database
Exports (Cash Basis) Overall Output (Value-Added) 1a. Papers SCI b. Patents Triadic c. Patent PCT Apps a. GERD b. BERD a. Researchers b. Business Researchers a. Size GDP b. Size Population Coefficients of determination (R 2 in %) of HT exports and overall HT manufacturing sales with explanatory variables in Uses log scales. BERD (Business Expenditures on R&D) has the strongest correlation, but overall GERD (Gross Expenditures on R&D) is not far behind. Papers and patents are also good predictors.
This is a very good fit, maybe too good! Regression model for overall high-technology manufacturing sales (value-added) vs. business expenditure on R&D, both in 2009.
Lags do not significantly improve models, even for the aerospace sub-sector. Likewise, adding a second explanatory variable does not significantly improve models. Lag analysis
The US and EU have linear trends, while the PRC curve is clearly quadratic. Forecasts of BERD using extrapolations. China will probably not take the lead until 2018.
HT manufacturing sales, with extrapolations after The PRC trend model is quadratic, and the others are linear. Value-added data. The high technology sector has been growing rapidly worldwide. China will probably take the lead in 2014.
Correlations are with (a) BERD in various years with HT manufacturing in 2009, and (b) HT manufacturing in various years with BERD in BERD as a result is somewhat more convincing. Use of lags to illuminate cause and effect in two variables.
Conclusions New value-added data on HT manufacturing provide more accurate trends than exports on a cash basis This data correlates very closely with business investment in R&D, which suggests governments should encourage it But, this may be spurious, since it may be an effect rather than a cause. That is, it seems that sales enable R&D investments
Appendix Paper posted at WTEC is at ITRI is at New value-added data is at aSetCode=TIVA_OECD_WTO
The loss of US share seems to be to the ones listed. Early US gain seems to have come from Russia