CAPITAL APPROPRIATIONS AND BUSINESS FIXED INVESTMENT IN MANUFACTURING: AN ECONOMETRIC ANALYSIS*

Abstract

THE PURPOSE OF THIS STUDY is to examine a model of business fixed-investment spending in the context of a fully specified econometric model of income determination. Investment decisions are measured by appropriations for capital spending rather than by capital expenditures themselves. Although data on appropriations are restricted to the 1,000 largest manufacturing firms, this important subset accounts for virtually all of the variation of capital spending in manufacturing and a substantial portion of the variation of all business fixed-investment expenditures. The relationship between expenditures and appropriations is treated separately from the relationship between appropriations and the determinants of investment decisions. Appropriations are assumed to be determined in an internal (to the firm) market for investment funds. This view incorporates both accelerator and internal-funds theories of capital spending into the theoretical structure used to explain investment decisions. The relationship between expenditures and appropriations is treated as a simple linear distributed lag. The resulting two-equation investment system was estimated by ordinary least squares using annual first differences for total manufacturing and several constituent manufacturing industries. The results indicated fairly general support for the theoretical structure. Expansion requirements (the difference between desired and actual capital stock) and the subjective or internal cost of funds consistently appeared as significant determinants of appropriations. The long-term interest rate was not generally a significant determinant of appropriations, but did appear important in some individual industries. The results of the distributed-lag estimations indicated that it takes about three years for a dollar of appropriations to be fully realized in expenditures, though well over half of each appropriation dollar is phased into expenditures within a year and a half of the formal commitment. A set of estimated equations for total manufacturing which included the appropriations equation, the distributed-lag equation, and several linking equations was introduced into the University of Michigan Econometric Model of the U.S. Economy. The Michigan model was then used to generate ex post forecasts of important measures of economic activity for the period 1960 through 1965. The results indicated that the incorporation of the estimated investment system substantially improved the forecasting accuracy of the full system when compared with forecasts which relied solely on businessmen's anticipations of capital spending for projections of business fixed investment. Even though a measure of capital appropriations was available for only the 1,000 largest manufacturing firms, the accuracy in the prediction of capital expenditures in this important subset of firms was sufficient to improve the overall forecasts of economic activity.