Estimation of potential GNP in the U.S. economy : an application of input-output analysis and alternative updating methods

Abstract

Previous approaches to measuring the annual potential gross national product (GNP) of the U.S. economy have suffered from several common shortcomings. For example, the demand for goods and services in the economy and hence prices and wage rates were explicitly ignored. Also ignored were the capacities of the economy's manufacturing sectors, including agriculture, and the interindustry relationships which exist in the economy. This study approaches the measurement of potential GNP in a general equilibrium context. Thus, the measure of potential GNP in this study will represent the nation's desired output when its potential labor force and/or its existing manufacturing capacity is fully employed. To accomplish this task, a series of annual quadratic input-output models was developed in this study covering the 1958-79 period. Because of the lack of published input-output transactions tables in the later stages of the time period covered by this study, a means of projecting the technical coefficients required by the annual quadratic input-output models for these years was needed. An extensive review of the literature on input-output updating techniques revealed a general lack of consideration given to the inclusion of economic factors suggested by production theory. A new updating technique was proposed whereby the matrix of technical coefficients associated with intermediate products and primary inputs were regressed upon appropriate sets of relative prices. A comparison of this method with the frequently-used RAS method showed that the method proposed in this study out-performed the RAS method in periods of changing economic conditions. The proposed technique was then used to project the technical coefficients for those years during the seventies in which these coefficients were unavailable. The estimation of potential GNP given by the series of annual quadratic input-output models developed in this study were shown to compare favorably with the published estimates made by the President's Council of Economic Advisors based largely upon Okun's law and by Clark's production function approach. For example, the GNP gaps implied by the estimates of potential GNP were shown to do a better job of explaining the inflationary pressures which existed during the time period covered by this study than either of the two previously mentioned sets of published estimates.