| This dissertation is aimed at evaluating the ability of three R&D-based growth models developed by P. Romer, Aghion and Howitt, and Grossman and Helpman to explain the actual growth process. Specifically, using US time series covering the period 1963–1999, and data on 13 OECD countries, the hypothesis that the sign of the relationship between the real interest rate and the share of labor employed in the research sector is consistent with what is predicted by the afore-mentioned models is tested. The statistical results support the exact opposite hypothesis, that the two variables are positively correlated. Given the inconsistency between the models and the data, I examine whether any of the models has the ability to explain, under different assumptions, the variation in the data.; An examination of all three models and of historical series of related variables leads to the following conclusions. (1) The models developed by Aghion and Howitt and Grossman and Helpman are unable to explain the positive correlation observed in the data. (2) Romer's model is qualitatively capable of explaining the behavior of the data if variations are driven by changes in research productivity or government subsidies to the research sector. (3) This hypothesis can be supported by historical facts. (4) An important determinant of the long-run real interest rate is the rate of technological growth. Therefore the real interest rate will be sensitive to changes in variables that affect the technological growth rate. The most important such variables are government subsidies to research as a share of GDP and the productivity of the R&D sector. (5) Developed economies with more productive research sectors might offer a higher rate of return to savings than developing countries, even if the latter are characterized by a lower capital to labor ratio. (6) The absence of scale effects in the data, which is considered the most significant weakness of R&D based growth models, can be explained by the increasing specialization in the sciences. This path of research seems more promising than assuming decreasing returns to knowledge or increasing varieties as has been suggested. (7) Further examination of the quantitative properties of the model and their ability to match the data is necessary. |
