Growth Accounting vs. Regression In a Cross-country TFP Exercise

Despite occasional objections total factor productivity (TFP) has been widely recognised as the key determining composite component in the literature on economic growth. Most empirical studies of the contribution of TFP to growth are based simply on the conventional calculations of growth accounting. This paper argues that judging TFP solely in terms of growth accounting carries risks and loses information, and suggests an integrated approach that uses the TFP estimates created from growth accounting as inputs into a panel data analysis. This allows a close investigation of TFP from a new and revealing angle. The results, using 1960-1997 data for 16 OECD countries, strongly support the view that TFP is by far the most important factor in explaining growth. This is consistent with the recent arguments of Klenow (1997), Prescott (1998) and Easterly and Levine (2001). Furthermore the results show that the linkage between labour input and growth, given the data on TFP, is stronger than that between capital and growth. The implication of these findings is that TFP, rather than the growth rate itself, should be the focus of attention when we try to understand economic performance. The approach developed here is used to re-examine the benchmark studies of Solow (1957), Christensen at el. (1980) and Jorgenson at el. (1987). It is shown that the results of their studies are consistent with those in this paper. JEL No. C23, D24, O40 Authors' emails: [email protected], [email protected]