Technological change, diffusion and output growth
The thesis presents a critical review of both traditional and new growth models emphasising their main implications and points of controversy. Three main research directions have been followed, refining hypothesis advanced in the sixties. We first find models which follow the learning by doing hypothesis and therefore consider knowledge embodied in physical capital. The second class of models incorporate knowledge within human capital while the third approach considers knowledge as generated by the research sector which sells designs to the manufacturing sector producing capital goods. A typical outcome of such models is the existence of externalities which causes divergence between market and socially optimal equilibria. Policy intervention aimed at subsidising either human capital or physical capital is thus justified. Empirical analysis has received new impetus from the theoretical debate. However, past empirical tests are mainly based on heterogeneous cross section data which take into account mean growth rates over given periods of time, and ignore pure time series analysis. On empirical grounds, the role of investment in the growth process has been emphasised. This variable has also been decomposed to consider the impact of machinery and equipment investment alone. In this thesis we have underlined six aspects of endogenous growth models, which in our opinion reflect the main points of controversy: i) scale effects; ii) the treatment of knowledge as a production input; iii) the role of institutions; iv) the empirical controversy dealing with the robustness of growth regression estimates and the measurement of the impact of some crucial variables (e.g., investment) on growth; v) the simplified representation of R&D; vi) the absence of any discussion of diffusion phenomena. We then propose a new version of an R&D endogenous growth model, which explicitly incorporates the diffusion of innovations and permits comparison with results derived from other models which do not consider the diffusion process. In this new model the interaction between the sector producing final output and the sector producing capital goods generates the time path of diffusion and hence the growth rate of the economy. In this new model there is a clear growth effect of a change in the interest rate. Such a change, on the one hand, affects the determination of the value of human capital in research, and, on the other hand, affects the diffusion path of new producer durables. This is important for policy because policy aimed at stimulating growth may be mainly concerned with reductions of the interest rate and will thus cause a higher allocation to human capital in research and a larger supply (and use) of new intermediate goods. In addition, there is another clear growth effect which derives from changes in the parameter which defines the diffusion path of new capital goods. An increase in the value of this parameter again causes an increase in human capital devoted to research and an upward shift of the diffusion path, thus increasing the long-run growth rate. This result underlines the difference with previous R&D endogenous growth models in that we now have a clear distinction between the sectors producing and using new capital goods. The empirical implications of the theoretical models are then investigated by testing the causal link between R&D and investment, on the one hand, and output growth and investment on the other hand. Indeed, a crucial task of any empirical investigation dealing with endogenous growth theories is to explain the nature of the links between industrial research, investment and economic growth. There is much room for study in this framework, as there are still only a few studies analysing these relationships. Our analysis deals with both aggregate data for the US and UK economies and an intersectoral analysis for the US manufacturing sector. We have used a test procedure which allows us to analyse both the short-run and the long-run properties of the variables using cointegration techniques. We are able to test for any feedback between these variables, thus giving more detailed and robust evidence on the forces underlying the growth process. The results suggests that R&D Granger causes investment in machinery and equipment only in the US economy. However, there is evidence of long-run feed-back implying that investment may also affect R&D. In the UK economy there is no evidence for R&D causing investment nor is there strong evidence of long-run feed-back between the two variables. This suggests that the causal link between R&D and investment may not be thought of as a stylised fact in industrialised economies. We have also analysed the relationship between investment and output growth to test whether investment may be considered as the key factor in the growth process. We find little support for the hypothesis that investment has a long-run effect on growth. In addition, causality tests support bi-directional causality between these variables in the US economy while in the UK economy, output growth causes investment both in the shortrun and in the long-run.