One of the most well-known empirical regularities in the R&D-productivity literature is the existence of substantial under-investment in R&D. This strongly suggests that government should actively promote research activities. However, the so-called`quality-ladders' models of endogenous technological progress are inconsistent with this observation. In an extreme case, Grossman and Helpman (1991, Ch.4) suggest that R&D should always be taxed irrespective of the size of quality improvement. This paper attempts to reconcile these empirical and theoretical ®ndings by showing that the normative results of Grossman and Helpman are not robust.

This paper contributes to the endogenous versus semi-endogenous growth debate by establishing that semi-endogenous growth is more general than endogenous growth in a two-R&Dsector growth model. It is demonstrated that endogenous growth requires two`knife-edge' conditions of parameters. This ®nding (i) is in sharp contrast to recent two-R&D-sector models that show that long-run growth is endogenous, and (ii) resurrects the policy conclusion of semi-endogenous growth that government policy is not effective in raising the underlying growth rate of an economy. The driving force of these results is knowledge spillovers between two R&D activities, which are largely neglected in existing studies.

We set out an infinite-horizon political economy model with partisan and office motivation effects in an endogenous growth context to demonstrate that the existence of political uncertainty regarding re-election tends to reduce the amount of public investment by incumbent governments and underlies a switch from government investment to government consumption, thereby reducing growth. The political equilibrium is inefficient and so does not maximise social welfare. Using panel data regressions we show, for OECD countries, that there is empirical support for the hypothesis that political uncertainty tends to reduce public investment, and that there are partisan effects in public investment decisions

am grateful to Philippe Aghion, Julia Darby and Elias Dinopoulos for their valuable comments. Needless to say, I am solely responsible for opinions and all remaining errors.