Measuring and Forecasting Consumption

Abstract

During the past decade, under the stimulus of changed economic conditions and changed attitudes toward policy formation, real progress has been made in the measurement of consumption in the United States. Concepts have been clarified considerably, techniques for gathering information have been improved, some information has been collected, and a certain amount of lay as well as professional support has been developed for further use of public funds to measure consumption. Work in this field, however, is still in a pioneering stage, particularly with reference to the measurement of changes in the physical volume of consumption.

One purpose for which information on consumption has been used increasingly during recent years is the forecasting of changes in consumption and in economic conditions generally. Consumption has been widely regarded as a variable dependent in quite regular fashion on certain other variables, particularly consumer income after taxes and increments therein. This interpretation has facilitated calculations of hypothetical figures for future gross national product and for future employment; and has contributed also to conclusions reached about prospective aggregate demand and aggregate supply, and prospects for inflation.

In the author's view, generalizations about the relationships between consumption and other elements in the economic situation have been made without adequate recognition of differences in behavior resulting from differences in time, place, and circumstance. This accounts in part for the shortcomings of many forecasts made last summer. Five suggestions concerning forecasting are offered for consideration: (1) that consumption in any particular period be estimated after close study of the prospects for component parts, with income after taxes being regarded as only one (very important) factor influencing consumption; (2) that more attention be given to accurate measurement of the current level of the physical volume of consumption as a starting point for estimating future changes in consumption; (3) that the common assumptions about constant prices be modified to suit the occasion and that the whole approach to the problem of estimating the probable course of consumption, income, employment, and prices be re-examined and revised to take adequate account of all the important factors in the market; (4) that very real limits to what can be accomplished in forecasting by describing the economic world in terms of mathematical relationships be recognized and that the importance of selective judgments with respect to particular periods be emphasized; and (5) that, in view of the present uncertain state of the art of forecasting, care should be taken in formulating policy to rely only so far as is necessary on forecasts arrived at by any method.