Energy consumption , technological progress and economic policy ∗

We rst provide an empirical study of the energy-saving technological progress on the ENERDATA database. Energy intensity is shown to decrease over the period 1971-1999 in OECD countries, indicating a signi cant energy-saving technical progress trend. We also show via semi-parametric partially linear estimations that: (i) this trend is positively correlated with the investment rate, and (ii) the marginal productivity of investment has accelerated in the sub-period 1985-1999 compared to 1971-1984. Second, we build a general equilibrium vintage capital model with embodied energy-saving technical progress to formalize these ndings. In this set-up, we study to which extent a steady increase in the marginal productivity of investment goods and/or scrapping subsidies could compensate the output loss due to a cut in energy use. The latter scal policy is shown to be particularly ine cient in this respect, specially under slow energy-saving technical progress. In the end, our model predicts that the implementation of Kyoto-like protocols in the computers age is much less painful than what it could have been two decades ago, particularly in the USA, which can even improve their relative position if the requested energy cut rates are not "too" unequal across countries.