Economic forecasters have a vast array of information to work with and a growing variety of techniques. A few economists, believing that just one or two key factors determine the future course of the economy, limit their observations to these factors and develop forecasts based on them. A leading example of this is found in the school of thought that ascribes most importance to changes in the money supply. But most economists use a wider range of material.

Some elements of the future are known with reasonable accuracy. Government spending is reflected in existing budgets. These budgets indicate how much will be spent and how much money will be extracted from the stream of private spending by taxation. Similar information is available on some parts of the private economy. Periodic surveys conducted both by government and by private organizations measure business plans to invest in new plants and equipment. Increasingly, attempts are made to probe the mood and intentions of consumers concerning the possible purchase of automobiles, houses, appliances, and other durable goods. Regular surveys are also made to determine the general mood of the public—whether people are optimistic or pessimistic about their own economic future and thus whether their spending is apt to be relatively strong or relatively weak. In general, such information obtained from the various surveys of investment plans, spending plans, and attitudes has been highly useful to economic forecasters. Such information helps to limit the range of possibility. But plans and attitudes change, sometimes quite abruptly, and although the surveys are useful tools they are not clear and reliable guides to the future.

Probably the single most difficult economic forecasting problem is to pick the turning points in economic activity—the times at which the economy turns from growth to recession or from recession to recovery. Because of the difficulty and importance of the problem, major efforts have been made to develop tools for this purpose. The National Bureau of Economic Research in the United States has identified a number of statistical series that normally turn up or down before the economy does. Common stock prices, business inventories, and changes in consumer installment debt are among these series, which are known as “leading indicators.” Other statistical series normally move in line with overall activity (“coincident indicators”), and a third group changes direction after the economy does (“lagging indicators”). Although careful study of these groupings of statistics can be helpful, unfortunately there is no fixed relation between the movements of the economy and those of the various indicators. Although the “leading indicators” do ordinarily lead, the length of the lead varies. This reflects the dynamic nature of a complex economy that is constantly changing and in which strength or weakness may come from a variety of sources.

Some economists also use sets of statistics called diffusion indexes to calculate economic turning points. A diffusion index is a method of summarizing the common tendency of a group of statistical series. If a greater number of the series are rising than are declining, the index will be above 50; if fewer are rising than declining, it will be below 50. In effect, a diffusion index measures the degree to which either strength or weakness pervades the economy. If, for example, most of a group of industries are increasing their production rates, the economy as a whole is probably expanding; if the proportion of industries that are growing begins to decline and falls significantly below 50 percent for a period of time, the economy is probably in a recession, or at least moving in that direction.

Economists frequently use mathematical equations to express the normal relations between various economic factors. As a simple example, a given increase in consumer income will ordinarily produce a certain increase in sales, saving, and tax revenue, and these developments can be expressed mathematically. With a sufficient number of equations, all the important interactions within the economy can be simulated in a mathematical model. With the advent of computers able to make millions of calculations in a few moments, economists began to construct more and more complex sets of equations, called econometric models. These models, some of which include hundreds of equations, can be used to forecast overall economic activity (macroeconomic forecasting) or developments in particular parts of the economy (microeconomic forecasting). The success of econometric forecasting has so far been limited because the exact nature of economic relations is not fully known, and also because of the inadequacies of existing statistics. Nevertheless, the improvement of these techniques represents the greatest hope for more accurate economic forecasting in the future.

The computer has also stimulated development of another potential forecasting tool, input–output analysis. Input–output tables show the relations between the various industries and sectors of an economy. They show, for each industry, the amount of its output that goes to every other industry to be used as raw materials or semifinished products, as well as the amount that goes to the final markets of the economy. Input–output tables also show each industry’s consumption of the products of other industries, as well as each industry’s contribution to the production process. With such a table it is possible to trace the effects of changes in one industry or sector upon all the other industries and sectors.

The usefulness of input–output analysis for forecasting purposes has been limited by a number of factors. One problem is that it takes years to put such complex sets of statistics together; in a changing economy, relations may have shifted by the time the data for a base period have been assembled. Progress has been made, however, in developing methods to bring these relations (called technical coefficients) up to date, and input–output analysis shows increasing promise as a forecasting tool.

Major improvements have been made in the accuracy of economic forecasting. A competent economist can usually predict accurately enough to provide guidance to those who make policy decisions. Consensus forecasts—the average of a group of forecasts made by different individuals or organizations—have come closer and closer to the mark in recent years. Errors persist, nevertheless, and they occasionally lead to bad decisions.

The sources of error in economic forecasts are many. Some lie outside the realm of economic analysis; wars, agricultural or other natural disasters, or political upheavals are examples. Some forecasts go wrong for essentially ideological reasons: people who do not believe that an economic system will function tend to forecast its failure, which accounts for the many predictions of another great depression. Adherents of a political party in power have a notable tendency to optimism, whereas their opponents, including economists, tend to view the future with alarm. The student of forecasts must obviously consider their source; purity of motive is an important virtue for economists.

The most vexing sources of error, however, lie within the realm of economic knowledge. Many are statistical. Not only are some of the published data inaccurate but even the best statistics are available only after a period of time; the forecaster is forever predicting the future when he cannot be completely sure of the present. Statistics of inventories, among the most volatile economic elements, are noteworthy in this respect.

The most persistent form of error in economic forecasting, however, is probably theoretical. Man’s knowledge of his own economic institutions is limited. Good analysis is made more difficult by the fact that these institutions are constantly changing. This means that economic theory based on experiences of the 1950s may be of limited use in the 1980s. Some of the greatest contributions to the continued improvement of economic forecasting may come from economists who are not necessarily forecasters themselves but have the insight to understand the changing economy of today.

Various of the economic indicators commonly used for forecasting are surveyed in Kenneth C. Land and Stephen H. Schneider (eds.), Forecasting in the Social and Natural Sciences (1987); John Llewellyn, Stephen Potter, and Lee Samuelson, Economic Forecasting and Policy—the International Dimension (1985); and Geoffrey H. Moore, Business Cycles, Inflation, and Forecasting, 2nd ed. (1983). Techniques for selecting appropriate data and models for use in economic forecasting are discussed in Giles Keating, The Production and Use of Economic Forecasts (1985); Lawrence R. Klein and Richard M. Young, An Introduction to Econometric Forecasting and Forecasting Models (1980); Steven C. Wheelwright and Spyros Makridakis, Forecasting Methods for Management, 4th ed. (1985); and Norman Frumkin, Tracking America’s Economy (1987). One of the most useful sources of timely statistical data for U.S. economic forecasts is Survey of Current Business (monthly).