Economists and the Future

Mr. Reed is Chairman of the Department of Economics at Northwood Institute in Midland, Michigan, and Director of the college’s annual Freedom Seminars.

In September, 1981, an economist from a major university in Michigan made known his economic forecast for 1982. His prognostications were widely publicized; perhaps some business or government decisions were based on them. According to the economist, the sluggish condi tions of 1981 would give way to recovery early in 1982. Auto sales would improve to an annual rate of 9.7 million vehicles by the second quarter. Unemployment would “stabilize” at about the 8 per cent level. Economic expansion would spread to all major sectors, with the overall “rate of growth” doubling by the end of the year. Price rises would remain about as strong as they were in 1981. Many of his predictions were expressed in precise mathematical quantities.

What a difference a few months can make! Anyone who was awake last year knows that this particular forecaster entirely missed the mark. And yet, he employed one of the most sophisticated mathematical models money can buy.

What does an economist do, having erred so grievously? Quietly retreat into the shadows of academe? Not at all! Undaunted, he will wipe the egg from his face, resume his place in the crowded fraternity of economic soothsayers, and begin work on next year’s prediction for Gross National Product—to the nearest tenth of a per cent. In the welter of fallacious forecasts, hardly a soul will single him out anyway.

The dismal record of the forecasting profession led one economics professor at the State University of New York to conclude recently that non-economists on balance are better at seeing the future than are the professional forecasters. Perhaps it is time for the professional soothsayers to re-examine their premises and methods.

What is it about the future that makes it so hard to describe? The answer is at once both simple and profound: it hasn’t happened yet!

Human hindsight is often “20-20" but it is beyond human mental limits to really know with much precision what tomorrow will bring. No palm reader, no fortune teller, no astrologer, no forecaster, not even an econometrician, can ever dispel the uncertainty of the future. Austrian economist Ludwig von Mises, in Human Action, tells us:


If it were possible to calculate the future state of the market, the future would not be uncertain. There would be neither entrepreneurial loss nor profit. What people expect from the economists is beyond the power of any mortal man.[1]

So it is that the existence of uncertainty is a commentary on the nature of the human condition itself. It is what Murray Rothbard terms “a fundamental implication derived from the existence of human action.” In his monumental work, Man, Economy, and State, Rothbard expounds:

This must be true because the contrary would completely negate the possibility of action. If man knew future events completely, he would never act, since no act of his could change the situation. Thus, the fact of action signifies that the future is uncertain to the actors. This uncertainty about future events stems from two basic sources: the unpredictability of human acts of choice and insufficient knowledge about natural phenomena. Man does not know enough about natural phenomena to predict all their future developments, and he cannot know the content of future human choices. All human choices are continually changing as a result of changing valuations and changing ideas about the most appropriate means of arriving at ends. This does not mean, of course, that people do not try their best to estimate future developments. Indeed, any actor, when employing means, estimates that he will thus arrive at his desired goal. But he never has certain knowledge of the future. All his actions are of necessity speculations based on his judgment of the course of future events. The omnipresence of uncertainty introduces the ever-present possibility of error in human action. The actor may find, after he has completed his action, that the means have been inappropriate to the attainment of his end.[2] (emphasis Rothbard’s)

There Is a Need to Judge What the Future May Bring

To say that the future is uncertain, however, does not mean the end of the matter. Surely, entrepreneurs who assemble the tools of production today, as Rothbard points out, must make decisions based upon what they think the future will hold. They make it their business to grapple with questions such as: What will the general state of business be next year? How much will materials cost and will they be available? What wage rate will be required to attract and keep the kind of employees we need? What will be the effect on sales if we change our prices? What are our competitors likely to do? Where will the best markets for our products be? Is this a good time to seek outside financing or will interest rates decline in coming months? Should we be working down our inventories? What will the politicians do that might affect our business?

Consumers, securities investors, government policymakers, and, of course, economics professors on the lecture circuit, join businessmen in the search for information about the future. The real question is, what can we reasonably say about tomorrow and what methods enable us to say it? A review of the more prominent methods of economic forecasting is now in order.

A. Simple Trend Projection

This approach relies upon pure extrapolation of previous trends in some economic activity and as such offers little more than a pretense to being scientific. It works only insofar as current trends continue. It does not begin to account for, let alone incorporate, any significant changes or turning points. Professor James B. Ramsey terms it “naive prediction” and offers this critique:

Either the predictor estimates some relationship and assumes that the same results will hold in the future; or he predicts values by using currently observed trends in economic variables over time, for example, he says next year’s income will be equal to this year’s plus 5 per cent. There is no attempt to provide a theoretical model in order to understand the observed relationships. There is no concern for identification and little for separating out the individual effects of exogenous variables.[3]

Thomas Malthus, early in the nineteenth century, used a kind of simple trend projection to forecast starvation and over-population. More recently, the so-called “Club of Rome” relied on the same approach to predict the same thing. In Malthus’ case, the Industrial Revolution interfered with his projection rather decisively. The Club of Rome’s projection did not foresee the decline of birth rates in industrialized countries.

To the extent that forecasters era-ploy simple trend projection (and many of them do), they are walking on ice so thin you can hear it cracking as they go.

B. Gross National Product Models

The concept of GNP purports to express the total value of all goods and services produced during a given period of time. It is the consummation of “national income accounting”—the process of identifying and adding up all the components which comprise the economy.

Basically, GNP is “determined” either by (a) summing the total expenditures on the “final product” goods and services produced during a period or (b) summing the total cost incurred as a result of producing the goods and services applied during the period.[4]

GNP is probably the most widely used “measure” of total economic activity and is the statistic which most conventional analysts use to express their predictions of business performance. Its many components supposedly, comprise a “model” of the economy which can be a foundation for economic forecasting.

What on the surface appears to be massively profound turns out to be something much less. GNP, being the most “aggregate” of statistical aggregates, is riddled with problems and errors and, what’s worse, problems and errors of unknown magnitude.

Those problems and errors stem from both the complexities of statistical measurement and the difficulties of basic conception (what to include). What follows is an accounting of just a few.

1. Errors of estimation. Simon Kuznets himself, the “father” of GNP, suggested once that assuming an average margin of error for national income estimates (a prime component of GNP) of about 10 per cent would be reasonable! Yet, some economists routinely predict quarterly GNP figures in tenths of one per cent. Congress often makes public policy based upon those compu tations which, even if accurate, conjure up what Roger Garrison describes as “the vision of a dietician who weighs a locomotive both before and after the crew boards it, then uses the difference between the two weighings as the basis for prescribing a diet for the whole crew.”[5]

2. Incentives for collectors of the data to fabricate or twist the statistics for personal or political advantage. We know that economic statisticians in communist and Third World countries are notorious for this. Is it really unreasonable to assume that some twists or fabrications happen here too? In a recent, rather blatant example, the government decided to quietly start counting the 1.7 million members of the armed forces in this country as part of the work force for the first time. That at least will make the official unemployment figures look better for those in public office.

3. Incentives for individuals providing the statistics to report incorrect figures. Such distortions occur as individuals attempt to guard trade secrets, evade taxes, or mislead competitors.

4. No account is made for the activities of the “subterranean economy.” Giving Caesar the slip has become common practice as Americans are called upon to dig deeper in their pockets for what Caesar claims is his. Underground transactions, which totally escape the tax and data collectors, probably amount to hundreds of billions of dollars and probably are rising.

5. Things not exchanged for dollars are not included. Paint your own house and the value of the work performed is not calculated by the statisticians; hire a painter and his wages become a part of GNP. Likewise, if a man divorces his wife and then hires her as a cook for $100 a week, GNP will increase by $5200 annually.

6. Government spending raises GNP. When government spends more, it diverts funds away from more efficient allocation by the market. One economist suggested—with some sincerity—that it might be more in line with reality if government expenditures were subtracted from GNP!

7. Inappropriate depreciation allowances. These are determined by often unrealistic assumptions underlying the tax laws. Inflation in recent years, for instance, has rendered depreciation allowances quite inadequate.

8. Changing quality of goods not reflected. GNP would not rise if an improvement in a product did not result in a higher price.

9. Exclusion of leisure. Leisure is very much an economic good (subjectively valued and incapable of quantification) and people often opt to “consume” more of it and to consume less of the more “traditional” goods and services.

10. Frequent revisions. This shortcoming is related to the first one cited above. GNP statistics are constantly subject to revision. Those adjustments are often significant and sometimes come months or years after the initial calculation. In short, by the time we have a statistic which we can reasonably assume is “final,” it may have long since lost any forecasting value, if indeed it had any in the first place.

Reliance on Gross National Product models as tools for accurate forecasting has repeatedly led economists astray. It seems that, at best, such models say something about the past, and nothing about the future. Professor Kenneth Boulding’s reference to GNP as “one of the great inventions of the twentieth century, probably almost as significant as the automobile,”[6] goes down as a grotesque exaggeration.

C. Econometrics

Many of the problems of simple trend projection and GNP models are present in the more sophisticated, heavily quantitative, econometric models. These constructs, which many once thought to be quite promising, often comprise hundreds of mathematical equations that purport to represent relationships among the major aspects of economic activity. Expensive, high-speed computers churn out the meticulous forecasts of the econometrician.

The record of these models has been dismal indeed. Mistakes in econometric forecasts have often been so bad that merely changing their signs from positive to negative or negative to positive would have put them significantly closer to the mark. Business Week for March 30, 1981 provides a case in point:

The big econometric models began signalling a downturn early in 1979 and construed the second-quarter dip as the onset of a potentially serious recession. After the third-quarter recovery, they kept betting that the next quarter would turn negative. Then, when last spring’s drop was already under way, they turned briefly optimistic until the worsening statistics convinced them that their initial pessimism had been correct. They were wrong once again, because the economy picked up during the summerand was still running strong at the end of the year. “They were not only consistently wrong, they constantly changed their forecasts in the wrong direction,” notes Stephen K. McNees, an economist at the Federal Reserve Bank of Boston whom the econometricians themselves rely on as an arbiter.[7]

These errors certainly do not occur because the practitioners of this method do not try. They are simply employing inappropriate assumptions—assumptions that if rejected would lead to the virtual termination of econometric models as we know them.

Economics as a science is best analyzed qualitatively, not quantitatively. There are no truly constant relationships in human action, which means that most of the relationships postulated in the equations of econometric models are invalid. “Garbage in, garbage out,” as they say in computerese.

Economists have acknowledged for decades that the function of the entrepreneur is to anticipate changes in the marketplace. Once the entrepreneur has made a decision, he then exposes his wealth and income by arranging factors of production in such manner that he may satisfy future consumer demand. If he anticipates correctly, he will earn entrepreneurial profits; if his judgments are wrong he will incur losses. Any number of variable and unforeseen elements may arise to affect the outcome: changes in fashion and technology, government policy, labor union activities, competition, prices, and even the weather. None of these elements is entirely predictable; none can be accurately determined by past performance. Attempts to mathematically estimate these elements in advance or to attach numerical significance to the subjective judgments of the entrepreneurs themselves are pure folly. They are doomed to suffer the failure which lies in gross simplicity and imprecision.

Not a Precise Measure

It is ironic that econometrics strives for the exactness of numbers and yet bogs down in static equations which necessarily cannot begin to account for all the relevant factors and their interrelationships. Economist Henry Hazlitt tells us that if a mathematical equation is not precise, it is worse than worthless; it is a fraud:

It gives our results a merely spurious precision. It gives an illusion of knowledge in place of the candid confession of ignorance, vagueness, or uncertainty which is the beginning of wisdom.[8]

Perhaps Mises said it best when he wrote:

The fundamental deficiency implied in every quantitative approach to economic problems consists in the neglect of the fact that there are no constant relations between what are called economic dimensions. There is neither constancy nor continuity in the valuation and in the formation of exchange ratios between various commodities. Every new datum brings about a reshuffling of the whole price structure.[9] (emphasis mine)

The equations of econometric models profess complexity, yet they really represent a feeble, simplistic, and futile effort to mirror the infinitely more complex network of human actions we call “the economy.” They fail to account for many unforeseen economic variables and make little effort to recognize the interaction between economic and noneconomic variables. Their static, impersonal, and aggregative approach leaves acting man out of the picture, replaced by lifeless equations of often dubious value. The one way they could be reliably predictive would be if people ceased changing and became robots; then the econometrician could “get a handle” on them.

One observer recently commented that to predict economic events, one must first predict political events. Unfortunately, there is much truth in that statement. Today, it is not enough to consider endogenous market forces when contemplating the future. One must reckon with the exogenous influence on the market of colossal, erratic government. Politicians and their bureaucratic foot soldiers throw their weight around like bulls in a china shop. Predicting the outcomes of the political process is like trying to forecast which vases the witless bulls will break next. Econometric models are incapable of foreseeing such events.

The failure of econometric forecasting should come as no surprise. But it would be surprising were its practitioners to admit failure.

D. Statistical Indicators

This approach utilizes measurements of economic activity which supposedly “lead,” “coincide with,” or “lag” the business cycle.

A list of leading indicators generally includes the money supply, housing permits, stock prices, raw materials prices, inventories, and corporate profits.

Roughly coincident indicators include industrial production, factory capacity, retail sales, and personal income.

Unemployment, bank rates on short-term business loans, labor cost per unit of output in manufacturing, and new capital appropriations are considered key lagging indicators.

Obviously, the group which is supposed to have the most predictive value is the group of leading indicators. The Commerce Department compiles the monthly “Composite Index of Leading Indicators,” a widely followed statistic. Just how reliable is it?

The index’s lead time in signalling the onset of recessions has ranged from four months to nearly two years, which makes it a shaky guide for anyone trying to plan for economic swings.

The index’s performance in calling the upturns is only marginally better. On several occasions, it has signalled booms or busts which never materialized.

Statistical indicators, regardless of their category, often have substantial inherent weaknesses. Many of those weaknesses are akin to those described above with regard to GNP, itself viewed as “roughly coincident” to the business cycle.

The Producer Price Index, for instance, measures changes in charges by firms that make goods. It is based largely on returns from sellers, who tend to report list prices. Not recorded are the many trades that take place at discounts or at premiums.

The Consumer Price Index is the most-watched “cost of living” figure. It assumes that families buy items in the same proportions as they did in the base year of 1972-73, even though changes in lifestyles have since taken place. For one thing, it seems that an increasing number of Americans today are keeping their cars longer than they did ten years ago, so the purchase of a new car carries much less weight in a family’s budget.

Also, the CPI vastly overweights average housing costs and does not take into account the fact that people tend to buy more of a substitute when the price rises on their first choice. They buy more chicken, for example, when beef prices go up.

Official figures on unemployment are an important factor in government planning. But the figures, based on household surveys, are deceiving. For example, some able-bodied people cannot get certain types of welfare unless they are actually looking for work, so they may facetiously tell survey takers that they are job hunting. They then become officially unemployed.

Assuming it possible to assemble accurate statistics which indicate what they are supposed to and do not require later revision, we might have a sketchy picture of where “the economy” was or perhaps where it presently is. But we still couldn’t say for certain, based on the figures, where it is heading.

Educated Speculation

Having said all that, it nonetheless stands to reason that if we are to be able to say anything at all about the economic future, we probably should know something about the economic present and past. That’s where reliable statistics might play some part, not as a basis for simple trend projection, but merely as descriptions of economic activity already behind us or underway. Even the finest and most accurate statistics, though, should only be ingredients in a more fundamental approach now to be examined. For want of a more descriptive title, I shall call it Educated Speculation.

This approach is characterized by the following:

1. A clear recognition of the uncertainty of the future with no “leaps of logic” or mindless extrapolations.

2. Careful use of only the most meaningful statistics, understanding all of the limitations of such aggregates discussed above. This implies a task of “de-aggregating” aggregates—of analyzing economic activity as it results from acting, decision-making, welfare-maximizing individuals.

3. A sound understanding of basic economic principles and of the political process.

4. A thorough grasp of the causes and consequences of the business cycle.

With these tools, an economist can proceed to say something about the future and have some reasonable grounds for saying it. He still must be wary, though, of how far he can go. Brian McAndrew, writing in the Cato Institute’s Policy Report for November 1981, clarifies this point:

If forecasters recognized the limitations of economic theory and empirical information, they would realize that the most an economist can hope to do is explain the likely consequences of different policies. An economist can show, for instance, that a minimum wage tends to cause unemployment because it alters supply and demand conditions in the labor market. An economist cannot say exactly when, where, and by how much unemployment will rise (i.e., he cannot forecast the unemployment rate), but he can say that if a minimum wage law is instituted unemployment will tend to increase. In addition, he can, by combining theory with empirical information, get a rough idea of the amount of unemployment caused by the minimum wage at different times in the past, but he cannot say what this amount will be in the future.[10]

The Austrian Theory

In this world of radical interventionism, correct business cycle theory is crucial to our ability to say anything about the future. Cycle theories abound, but the one which fully integrates an explanation of the cycle and its features with an analysis of the entire economic system is known in various circles as the “Austrian malinvestment theory.”

Propounded first by Ludwig von Mises and later enlarged by Nobel laureate Friedrich von Hayek, the Austrian theory holds that the source of the cycle lies in money and credit expansion orchestrated by central authorities and proceeds to explain its effects. It is the theory which en abled Mises during the subtle inflation of the 1920s to warn of a coming depression. Few believed him until it happened. I direct the interested reader to more detailed accounts found in the works of Mises, Hayek, and Rothbard.

In the final analysis, the art of entrepreneurship is the art of “educated speculation.” It is upon the shoulders of the entrepreneur in the market economy that the burden of “educated speculation” rests. For him, it is, in the words of Rothbard, “a matter of intuition, ‘hunch,’ and deep insight into the slice of the market that the entrepreneur knows and is dealing with.”[11] Entrepreneurship remains a vital, creative talent which economists would do well to spend more time examining. (See two works by Israel M. Kirzner: Competition and Entrepreneurship and Perception, Opportunity, and Profit.)

“Educated speculation,” as I have termed it, is really economics brought down to earth. It may not be as fancy as econometrics or GNP modeling, but neither is it as pretentious. It says simply that an economist should be an economist, not an aspiring prophet.

The reader who began this essay hoping to discover a crystal ball may be disappointed that I have really offered nothing of the kind. Instead, what I have attempted to show is that much of what is commonly referred to today as “economic forecasting” goes far beyond the real abilities of economists to predict the future. Rothbard offers us this sobering reflection:

As Ludwig von Mises used to point out to those who were tempted to succumb to the razzle- dazzle of economic forecasting: If someone were really able to forecast the economic future, he wouldn’t be wasting his time putting out market letters or econometric models. He’d be busy making several trillion dollars forecasting the stock and commodity markets. Let it be a reminder to anyone tempted to partake of, or give credence to, this modern form of soothsaying.[12]


1.   Ludwig von Mises, Human Action: A Treatise on Economics (3rd revised ed.; Chicago: Henry Regnery Company, 1966), p. 871.

2.   Murray Rothbard, Man, Economy, and State (Los Angeles: Nash Publishing, 1970), pp. 5-6.

3.   James B. Ramsey, Economic Forecasting—Models or Markets?, Cato Paper No. 10 (San Francisco: Cato Institute, 1980), pp. 37-38.

4.   James D. Gwartney and Richard Stroup, Economics: Private and Public Choice (2nd ed.; New York: Academic Press, 1980), p. 116.

5.   See p. xii of Garrison’s Foreword to National Income Statistics by Oskar Morgenstern, Cato Paper No. 15 (San Francisco: Cato Institute, 1979).

6.   Gwartney and Stroup, p. 128.

7.   "Where the Big Econometric Models Go Wrong,” Business Week, March 30, 1981, p. 70.

8.   Henry Hazlitt, The Failure of the “New Economics” (Princeton, N.J.: D. Van Nostrand Company, Inc., 1959), p. 99.

9.   Mises, p. 118.

10.   Brian McAndrew, “The Failure of Econometric Forecasting,” Policy Report, November 1981, p. 6.

11.   See p. xi of Rothbard’s Foreword to Ram-sey’s Economic Forecasting.

12.   Ibid., p. xii.

{{article.Topic.Topic}} {{article.Topic.Topic}}

{{article.Title}}

{{article.BodyText}}