Abstract
The economic growth of nations is strongly associated with rising R&D expenditures. The present study therefore endeavours to examine the stylized patterns of evolution of R&D expenditures (R&D expenditures as a share of GDP) along the development course in 18 OECD countries for the period 1981–2012. Empirical findings of feasible generalized least square model highlight that as economic growth proceeds, R&D as a share of GDP also shows a rise in OECD nations. The study also analyzes the long-term growth trends and structure of research and development (R&D) expenditures in OECD nations for the period 1981–2012. Further, the study analyzes another important issue, namely whether OECD economies are converging/diverging in their R&D activities over the course of time. To this end, σ-convergence has been applied to examine this issue. Findings of σ-convergence highlight declining coefficient of variation (convergence) in R&D intensities as well as R&D expenditure in OECD countries for 1981–2012.
Notes
1 Chenery (Citation1960, 635) also used value added in manufacturing as a share of GDP as a dependent variable in his analysis where R2 is around 0.64 only compared to R2 0.96 where he used values added per capita in the manufacturing sector as a dependent variable. This is because the variance of the dependent variable in the former is low.
2 While analyzing the relationship between economic growth and structural change in R&D expenditure, some studies used R&D as a share of GDP as a dependent variable, whereas others, such as those of Teitel (Citation1987 and Citation1994a; Citation1994b) used R&D expenditure as a dependent variable. This substitution can have the effect of increasing the proportion of variance explained (compared to the shares used).
3 Chenery and Taylor (Citation1968), however, used sectoral value added per capita as a dependent variable for analyzing sectoral growth patterns in large, small-primary-oriented and small-manufacturing-oriented countries.
4 Accepting the validity of equation (2) allows us to calculate that particular level of income per capita (GDPPC) at which, ceteris paribus, the share of R&D in GDP reaches a maximum level.Taking first order partial derivative of w.r.t we get Thus, substituting the values of regression coefficients from and equation (2) in equation (i) above, we get .
Taking antilog, we get = $32359 ($32.3 thousand). This should be a maximum point since second partial derivative is less than zero, i.e. it is negative as shown below in equation (ii).
Substituting the values of coefficients from and equation 2, we get: