On the measurement of technological progress across countries

We construct 14 alternative measures of technological progress for 19 OECD countries over the period 1970–2000, distinguishing between measures of productivity gains actually obtained in a given country (TFP growth, Malmquist index) and technological progress at the world technology frontier (potential TFP growth, the “frontier shift” index). We then compare these measures according to a range of characteristics, shedding light on some of their relative weaknesses and strengths. We find that these characteristics are sensitive to the precision of estimates of the world technology frontier, and then we demonstrate that this precision can be increased substantially by allowing for imperfect substitutability between unskilled and skilled labor and using US state-level data apart from cross-country data for estimating the world technology frontier. Because none of the 14 measures dominates all others on all dimensions, we conclude that the choice of appropriate measurement method should be suited to the question addressed in each particular study.