Obsolescence, Embodiment, and the Explanation of Productivity Change

There have been two major approaches to the explanation of changes in total factor productivity or of shifts in aggregate functions. Solow [17; 18; 19] asked .. . whether technical progress is predominantly or disembodied, whether gross investment is or is not the major vehicle by which new knowledge makes its way into the process of production [19, 42-43]. To this question Jorgenson [10] and Hall [9] provided the devastating answer that for Solow's aggregate functions one cannot distinguish from disembodied change. Jorgenson and Griliches [11; 12; 13] shifted the focus of discussion to more direct, detailed corrections of measures of factor inputs. Their hypothesis and initial findings were that, after such corrections, total productivity change in the aggregate function is negligible. Their estimates, have been questioned strongly by Denison [5; 6]. In response, they have revised drastically their initial conclusion on the magnitude of total productivity change, and their most recent estimates imply an annual increase of total productivity of about 1.1 percent. This may be compared to Denison's estimate of 1.4 percent and to their own initial value of only 0.1 percent. As for Solow's basic question, they have consistently found that embodiment (or, in the authors' terminology embodied capital diminishing errors) plays a minor rolel [11, 247]. While all the above-mentioned papers make significant contributions to conceptual issues, or introduce valuable data, we find ourselves, once again, with sizeable measures of unexplained rises in total factor productivity. In this paper we derive new estimates of embodiment for the previously rejected Solow-Cobb-Douglas approach. In conflict with the results of Jorgenson and Griliches, we find that embodiment is indeed a very important aspect of productivity change. The distinctive analytical feature of the paper is that we base our estimates of embodiment on measures of obsolescence. First, we estimate depreciation rates from expected economic lives as revealed by studies of actual business practice. Second, we establish plausible boundaries, based on appraisal studies, expert opinion, and information on maintenance, for the relative magnitude of * Acknowledgements are due to Solomon Fabricant and to a referee of this Journal for helpful suggestions. The data used were developed under the auspices of the former Office of Business Economics U. S. Department of Commerce, in connection with the Interagency Study of Economic Growth. These data, however, are not official series. 1 Their allowance for changes in the quality of capital goods is made by way of minor downward revisions in the rates of growth of capital goods price indexes [11].