The Partition of Aggregate Investment

Abstract

The current fashions, theoretical conventions and empirical methods adopted by economists and econometricians change quite slowly. Twenty–five years ago, there was an active theoretical and empirical debate on suitable means for the partitioning of aggregate investment into expansionary (or net) and replacement components — with; with replacement investment being generally treated as a passive process that could be adequately represented by a simple proportionality hypothesis, while expansionary investment was linked to various notions of desired capital stock and some distributed-lag adjustments. The acceptance of this particular approach meant that the direct impacts of governmental policies, the costs of financing acquisitions, and incentive programmes were limited to net investment since replacement was unresponsive to fiscal and other stimuli, except indirectly through the gradual accumulation of successive annual vintages in our capital stock. Thus aggregate levels of capital consumption (depreciation, or replacement) were seen as quite unrelated to such incentives, to expenditures on maintenance and repairs, to changes in patterns of final demand, to strategic adjustments by large firms, and to a host of potential factors (including intertemporal aspects of technology and dynamic competition) that otherwise might have been considered quite influential. The constant integration of this simple partition in economists’ theoretical models and in many practical formulae for economic data over the last 25 years has made life much easier for both the governmental statisticians, who relied on less information, and academic economists, who could thus develop (seemingly) dynamic models which were free from the technical hazards associated with non-passive replacement. A convenient illustration for showing the persistence of this approach is readily obtained by comparing the contents of Jorgenson (1989) with any of this particular author’s econometric studies of investment from the 1960s and 1970s. (See, for example, Jorgenson [1963, 1965, 1974] or Hall and Jorgenson [1967].)