Investment-specific technological change and growth accounting

Greenwood, Hercowitz and Krusell have claimed that the Jorgenson form of growth accounting is conceptually flawed and severely understates the role of technological progress embodied in new capital goods ('embodiment') in explaining US growth. To the contrary, in this paper it is shown that in its technology aspects their model is a special case of the Jorgensonian growth-accounting model. What they call investment-specific technological change is shown to be closely related to the more familiar concept of TFP growth: statements about the one can be translated into statements about the other. Empirically, they claim that the proportion of US growth accounted for by embodiment is about twice as large as estimated by conventional growth accounting. But the difference between these estimates is found to be due more to data than to methodology.