On the measurement of technological progress across countries

The article considers 22 alternative empirical measures of country-level “technological progress”, computed for 19 highly developed OECD countries over the period 1970–2000 based on (i) the neoclassical growth accounting approach that adopts the Cobb-Douglas production function specification, (ii) a nonparametric approach where the world technology frontier (WTF) is constructed with Data Envelopment Analysis, and (iii) a “hybrid” approach that combines the two. Measures of TFP growth (capturing all output gains actually obtained in a given country that cannot be traced back to factor accumulation) are carefully distinguished from measures of technical change (capturing only technological progress shifting the WTF). Empirical properties of all 22 measures are compared according to a range of characteristics frequently discussed in the macroeconomic literature. The conclusion is that the choice of appropriate measurement methodology should be suited to the question addressed in each specific study, the simple growth accounting approach is generally insufficient, and particular attention should be paid to the empirical treatment of technical efficiency changes. The results are also sensitive to the precision of WTF estimation.