The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
OF PHILADELPHIA SOCIAL MAN AND THE NEW STATIONARY STATE H O W WELL D O ECONOM ISTS FORECAST? < PS H Q H MAY 1971 Social Man and the New Stationary State . . . Is halting the hectic pace of economic growth an answer to the problems we, as a society, confront? How Well Do Economists Forecast? . . . A careful evaluation of past successes and failures in forecasting suggests some surprises about this year's predictions may be in store. BUSINESS REVIEW is produced in the Department of Research. Ronald B. Williams is Art Director. The authors will be glad to receive comments on their articles. Requests for additional copies should be addressed to Public Services, Federal Reserve Bank of Philadelphia, Philadelphia, Pennsylvania 19101. FEDERAL RESERVE BANK OF PHILADELPHIA Social Man and the New Stationary State* by David P. Eastburn, President, Federal Reserve Bank of Philadelphia one's standard of living. Social man faithfully lugs his used Coke bottles to the neighbor hood collection center every Saturday morn ing for recycling; Economic Man is inclined to feel that if empty bottles were worth any thing, somebody would pay for them. My point is not that Social Man is soft hearted and -headed or that Economic Man is a Scrooge, but that both have much to learn from each other. I have developed this theme in a general way elsewhere**, but would like to explore one important aspect of it here. To introduce this aspect, let me read an excerpt from the writings of a distinguished philosopher: There are two kinds of people in the world today: those who are primarily concerned with making a living and those who are pri marily concerned with living with their fel lows. The first might be called Economic Man; the second, Social Man. One of our main problems these days is that Social Man and Economic Man see the same things differently. Social Man urged his Congressman to oppose the SST because it would make too much noise and might induce skin cancer; Economic Man, if he opposed the SST at all, did so because it would not get off the ground financially. Social Man sees the moon shots as an extrava gant use of resources at a time when much of the world is starving; Economic Man sees them as spinning off many scientific by products that some day can improve every Towards what ultimate point is society tending by its industrial progress? . . . . . . the tone and tendency . . . [of the speculations of political economists of the last two generations] goes com pletely to identify all that is economi cally desirable with the progressive state, and with that alone. *An address given before the 1971 Annual Meeting of the National Assembly for Social Policy and De velopment, Inc., Denver, Colorado, April 30,1971. **"Economic Man vs. Social Man/' Business Review, Federal Reserve Bank of Philadelphia, October, 1970. © 1970 by the New York Times Company. 3 MAY 1971 BUSINESS REVIEW we have become slaves to technology. This is missing because it was written a century ago. The author is John Stuart Mill, a genius who began learning Greek at the age of three and ultimately became the outstanding economist of his age. One of his most intriguing contributions is the concept of the Stationary State. This concept is now capti vating many who, overwhelmed by the prob lems confronting society, wonder whether the solution might be simply to stop this hectic pace of change and growth. It is now fashionable to talk of a new Stationary State. What are the chances that a new Station ary State is the solution to our social prob lems? Let us look at this question from the points of view of both Economic Man and Social Man. I cannot . . . regard the stationary state of capital and wealth with the unaffected aversion so generally mani fested towards it by political economists of the old school. . . . I confess I am not charmed with the ideal of life held out by those who think that the normal state of human beings is that of strug gling to get on; that the trampling, crushing, elbowing and treading on each other's heels, which form the existing type of social life, are the most desirable lot of human kind, or any thing but the disagreeable symptoms of one of the phases of industrial prog ress. . . . the life of the whole of one sex is devoted to dollar-hunting and of the other to breeding dollar-hunters. RECIPE FOR GROWTH There is room in the world, no doubt . . . for a great increase of population.. . . . But even if innocuous, I confess I see very little reason for desiring it. . . . [There is not] much satisfaction in con templating the world with nothing left to the spontaneous activity of nature . . . with . . . scarcely a place left where a wild shrub or flower could grow with out being eradicated as a weed in the name of improved agriculture. If the earth must lose that great portion of its pleasantries which it owes to things that the unlimited increase of wealth and population would extripate from it, for the mere purpose of enabling it to support a larger, but not a better or a happier population, I sincerely hope, for the sake of posterity that they will be content to be stationary, long before necessity compels them to it. Economic Man has developed a most sophisticated recipe for growth, and for the past century or so has been proudly serving up his product for us all to consume. The recipe relies heavily on the ingredients of science and technology. Scientists must be free to indulge their limitless curiosity about how nature works. Technologists must be free to apply the fruits of science to the satisfaction of human wants. People must want things and be willing to work to get them. Producers must be free to turn out these things with an ever-decreasing input of human effort— that is, with ever-increas ing productivity. Economic Man may say that he attaches no moral connotations to the result. It is neither "good" nor "bad"; is simply what people want, and if people want to risk cancer by smoking cigarettes, that is their decision; the system is neutral. Actually, Economic Man is seldom content with that position. He really believes that the freedom which the system requires is "good," and that the system has tremendously improved the physical well-being of mankind and that is "good." It is all there: a challenge to the unthinking assumption that growth, per se, is good; a denunciation of money-grubbing and over consuming; a clear brief for the zero popu lation advocates; and an eloquent appeal for preservation of the environment. The only thing missing is a complaint that 4 FEDERAL RESERVE BANK OF PHILADELPHIA PROOF OF THE PUDDING questioning many long-unquestioned as sumptions. In one of his columns in the New York Times, Leonard Silk has written about J. K. Galbraith: It is true that we have made tremendous advances in overcoming the niggardliness of nature and that the advances have been pro ceeding at a faster and faster pace. All of us have been made aware of this by a barrage of gee-whiz statistics in current literature. For example, Alvin Toffler in Future Shock calculates that The real problem today, says Mr. Gal braith, is not how fast we can increase production and consumption but how well— how happily—we are to live. 'In a rational life style, some people could find contentment working moder ately and then sitting by the street—and talking, thinking, drawing, painting, scribbling or making love in a suitably discreet way. . . . None of these requires an expanding economy.' Some other reactions, however, are not so healthy. One has been to seek refuge in nostalgia. In Philadelphia we have a new magazine whose journalistic precept (and I am quoting from its own advertising) is that . . . the child reaching teen age . . . is literally surrounded by twice as much of everything newly man-made as his par ents were at the time he was an infant. . . . Since the increases are com pounded, . . . by the time the individual reaches old age the society around him will be producing thirty-two times as much as when he was born. It is also true— in the United States, at least— that a remarkably large proportion of the people have participated in the growing affluence. To some extent, this is because Social Man has made a conscious effort to redistribute slices of the income pie, but to a great extent also it is because Economic Man has done such a good job of enlarging the total size of the pie. We all owe a great debt to Economic Man, and Social Man should not forget it. But Social Man is not so coy about at taching values to economic growth and, in his eyes, they are not all "good" ones. For most of the past century, he has been striv ing to round off the rough corners of eco nomic growth. His gadfly efforts have contributed child labor laws, pure food and drugs, the progressive income tax, social security, and countless other reforms. Still he is not content. Indeed, he feels he is losing ground. Mill's idea of the Stationary State, therefore, now looks like the ultimate solution to the mounting problems of eco nomic change. . . . too much attention is being directed to the counter-productive elements . . . in our society . . . [The] cover and graphics reflect the feel of a special quality, vintage nostalgia, wherein we can more comfortably explore the con structive workings of our people and community . . . nostalgia that reminds us that despite the problems and com plications of everyday living, life can still be romantic . . . . Another reaction has been fear and un certainty, fear of change and uncertainty in our ability to control it. J. Irwin Miller, a leading businessman and spokesman for the Committee for Economic Development— probably the most enlightened group of Economic Men in the United States— re cently wrote an article entitled "Can We Afford Tomorrow?" He began by saying that no earlier generation of Americans would have asked this. Such a reaction might have been expected. Ten years ago the most contagious idea around was that the age-old problem of REACTION TO CHANGE This is a healthy reaction. It involves 5 MAY 1971 BUSINESS REVIEW human frustration on the liability side? It is time to put a halt to growth, especially now that reputable scientists say growth is going to come to an end soon anyway. With our minds freed from the compulsion to increase material wealth, to consume more and more things, we can turn the full force of our attention to the social problems threatening to overwhelm us. This, I think, is what Social Man is saying, and I believe he is wrong. scarcity had been licked. Galbraith planted the seed with The Affluent Society. Expecta tions, already rising for other reasons, were inflated still further. Now we see that scarcity was not dead after all; while we were looking the other way, it again has reared its ugly head, more fearsome than ever. No wonder people are pessimistic. Recently, the president of the prestigious American Association for the Advancement of Science gave an address in which he was asking whether the ever-faster pace of growth, based on scientific discoveries, could continue. This is what he concluded: . . . so awesome is already the ac celerating rate of our scientific and technological advance that simple ex trapolation of the exponential curves shows unmistakably that we have at most a generation or two before prog ress must cease, whether because the world's population becomes insuffer ably dense, or because we exhaust the possible sources of physical energy or deplete some irreplaceable resource, or because, most likely of all, we pollute our environment to toxic, irremediable limits . . . . POSSIBLE COURSES Consider these possible courses for society to follow. Which promises greatest success in solving our social problems? Course #1 is to slow the rate of economic change. Population growth would be zero; the increase in consumption of gadgets would be cut back markedly; resources would be diverted to cleaning up the envi ronment, rebuilding the cities, upgrading education, and otherwise improving the quality of life. Course # 2 is to encourage economic growth and change. Economic Man would be given incentives to use technology to raise productivity, generating a rapid in crease in incomes. Out of these larger in comes, more would be made available for social and environmental use. Course #1 may have more idealistic ap peal, but it is not very practical. We talk about reordering priorities, but doing it is something else again. I find it hard to see the average American sitting still while a slowergrowing pie is being sliced up in a radically different way. He would, of course, benefit from clean water and no ghettos, but the benefit will be hard to see and touch. Per haps he can be educated to equate a clear stream with a new car, but this would be a long process. In any case, Course #1 would take time; for example, even if every couple started tomorrow to limit its family to two children, the population would not level off for another 66 years. . . . in future histories of the world the decade of the 1960's may be known . . . as the time when man, with un bridled lust for power over nature and for a so-called higher standard of living measured by the consumption of the products of an industrial civilization, set in motion the final speedy, inexo rable rush toward the end of progress. In short, the reaction of Social Man to economic growth is that he's been had. Eco nomic Man, he feels, may be a superb producer but a miserable accountant. What kind of an accounting system adds up on the asset side all the wonderful gadgets which the system has turned out but ignores pollution, exhaustion of resources, and 6 FEDERAL RESERVE BANK OF PHILADELPHIA must be recognized, however, is that the re sults of his efforts can be "bad" as well as "good," and that some means must be found to anticipate these results and channel these efforts. Course # 2 promises more immediate re sults. It would follow more easily from the path we have been taking; it would be better accepted because it would provide a bigger pie to be sliced up; and it would make greater use of the productive talents of Economic Man. "But," Social Man would say, "how can we be sure this would not lead to even a worse mess than we have now?" Unfortunately, there is no good answer. Neither Course #1 nor # 2 , therefore, is acceptable alone. What we need is a course of action that combines the best of both. I shall call this Course # 3. We should con tinue to encourage rapid growth but grad ually adopt some of those aspects of the Stationary State which are so appealing to Social Man. Let me suggest three important require ments for Course # 3 : 3. Planning. This will require more plan ning than Economic Man and probably less planning than Social Man would like. It probably means more Government inter vention. I see no alternative if rapid growth is not to continue its collision course with pollution, exhaustion of resources, and over population. The proper emphasis should be not on deliberate slowing of growth but on delib erate planning, for planning may tend to slow growth anyway. For example, the presi dent of American Cyanamid recently re ported that . . . 15 years ago we would have been happy to develop a plastic container that could be profitably marketed because it provided a safe and more convenient package for food or beverage. Today, we must inquire into the plastic's dis posability. And if the answer is unsatis factory, we had better go back to the test tubes.. . . 1. Incentives. Economic Man should be encouraged to do his thing. To the extent a free and competitive marketplace ac complishes this, fine. To the extent Govern ment must provide the incentive, this should be done. And to the extent "his thing" may produce undesired social results, incentives should be provided to lead him in the right direction. I have in mind, for example, tax incentives to install anti-pollution devices. This route can best channel Economic Man's energies in directions which Social Man desires. Family planning will tend to slow down economic growth, although not so much as businessmen might think. Increases in popu lation are not so important to a growing economy as some other things, particularly technology. And in any case, emphasis should shift away from overall growth to growth per capita; this is a much more sensi ble measure of well-being. Here, too, income per capita need not suffer significantly, and could well benefit, from a slower population growth. Finally, planning can help to stabilize the economy. Rapid growth will be increas ingly unacceptable if it is periodically inter rupted by severe recessions. Stop-and-go growth can be the most wasteful use of 2. Science and Technology. It is foolish to talk as if we can turn back the clock. A few philosophers may see the results of science and technology as dehumanizing, but most people see them as a miracle that has spared them a great deal of back-breaking labor. In any case, once a scientist has set his curiosity going, I doubt if anyone can turn it off. We shall need all the ingenuity he can bring to bear if we are to develop a cleaner automo bile engine or a quieter jet engine. What 7 MAY 1971 BUSINESS REVIEW A century ago Mill had a visionary idea. He concluded that: resources, human and physical. But to pre vent severe waste, it may be necessary when the economy is going too fast to slow it down deliberately for a while in order to sustain growth in the longer run. A stationary condition of capital and population implies no stationary state of human improvement. There would be as much scope as ever for all kinds of mental culture, and moral and social progress; as much room for improving the Art of Living, and much more likeli hood of its being improved, when minds cease to be engrossed by the art of getting on. CONCLUSIONS Course # 3 , with reliance on incentives and technology to stimulate growth but also on planning to channel it into socially ac ceptable directions, is more likely to lead us to the Promised Land than is pursuit of the new Stationary State. Course # 3 is a logical next step in the evolution of man's efforts to control his destiny. When Economic Man first applied technology to the satisfaction of human wants, he was rebelling against the blind laws of nature. Social Man wants to take the process one step further; he is rebelling against the blind forces of growth. In my opinion, this is something to work towards, but something we are not ready for. Now we need economic growth, and it would be a mistake to try to stop it. But we need to work towards better control of growth. Hopefully, before another century has passed, we shall have grown up to Mill's vision of the Art of Living. ■ N O W AVAILABLE The Federal Reserve Bank of Philadelphia annually publishes a Report of Operating Ratios. This Report contains average figures for member banks in the Third Federal Reserve District. To obtain copies of the 1970 Report, address your requests to De partment of Research, Statistics Section, Federal Reserve Bank of Philadelphia, Phila delphia, Pennsylvania 19101. 8 FEDERAL RESERVE BANK OF PHILADELPHIA How Well Do Economists Forecast? by Ira Kaminow might be forced to watch sales go down the drain because he failed to prepare for boom ing demand. If our national policymakers follow the wrong prediction, they may spend all their time locking the front door against unemployment while inflation quietly slips in through the back. With all this riding on the quality of our economic foresight, we should learn how to squeeze as much in formation as possible from predictions. Some of the most helpful guides to the intelligent interpretation of economic fore casts can be found by taking a hard look at past successes and failures. Only a search of actual experience can give us answers to those important questions that have been asked once again in the wake of the offi cial GNP forecast for 1971. Who has seen the Administration take up its lonely vigil at ten-sixty-five and not wondered how often victory in these things has gone to the maverick, or how often the forecasting fra ternity has underestimated the economy's potential? But experience also has lessons to teach that are only incidental to the ten- Not long ago, economists, businessmen, and the general public looked on in com plete surprise as the Nixon Administration forecast 1971 Gross National Product to be $1,065 billion. This projection was a good $15 billion more optimistic than nearly every other forecast around. The "ten-sixty-five" forecast loosed a barrage of attacks and counterattacks as economists began to man the battlements at every point between $1,035 billion and the "official" forecast. Most took up positions around the "stan dard" forecast somewhere between tenforty-five and ten-fifty. The names of the victors will not be known for many months after the end of 1971 when the final GNP figures are in. But bets must be placed now, by national policy makers, by businessmen, and by house holders. Millions of decisions will rest on which forecasters we trust and how much. The stakes are high. An individual who bets on the wrong man might wind up with a new car payment book in one hand and a layoff notice in the other. A businessman 9 MAY 1971 BUSINESS REVIEW growth. One man's "sluggish" is often an other's "rapid." Second, some forecasters were too irregular in their predictions to allow systematic evaluation. After eliminating all predictors whose forecasts were nonquantitative or were made sporadically, the list melted down to 13 forecasters. Of these seers, two were individuals or coordinated teams that made forecasts with substantial use of mathematical or econometric models’ of the economy; nine were individuals or coordinated teams that made forecasts con centrating on more traditional, less mathe matical techniques; and two were forecasters whose predictions reflect a consensus of separate, independently-arrived-at forecasts1 2 (see box). sixty-five debate. For example, it can tell us about the success (or failure) of the new econometric forecasting methods and whether any practitioners stand out as being the leaders in their trade. After we give history a chance to speak, we can assess more intelligently the meaning of economic predictions, especially those for 1971. TRAILING FORECASTERS Fortunately, forecasters have left behind a trail of published predictions that can be used to evaluate their past successes and failures. Each year since 1958, the Federal Re serve Bank of Philadelphia has recorded this trail by compiling a list of about 50 published forecasts for the coming year. These compila tions provide the raw material for a look at the predictive accuracy of economists. Unfortunately, not all forecasts, listed could be evaluated. First, all non-quantitative forecasts had to be dropped. A forecast of "sluggish growth" does not leave much room for objective comparison with actual 1 For a complete discussion of forecasting with econometric models, see Ira Kaminow, "A Nonecono mist's Nonmathematical Guide to Econometric Fore casting," Business Review, Federal Reserve Bank of Philadelphia, October, 1970. 2 The Fed compilation was not always complete and not all the published predictions of the 13 forecasters were recorded. The gaps were filled with information from a variety of sources. Traditional Tech nique. What we call the traditional technique is really any inde pendent forecast that does not rely heavily on a single econometric model. Tradi tional forecasters use a wide variety of inputs, such as pure intuition, leading indicators, trend projection, and econo metrics, to come up with pre dictions. Consensus. Some econo mists and organizations sim ply poll other economists on next year's outlook. The aver age of these forecasts is the "consensus" of the polled economists. MORE A B O U T M ETHODS O F FO R ECASTIN G Econometric Technique. Econom etric models are mathematical-statistical mod els of the economy. Some are used as tools in forecasting. However, contrary to some popular opinion, few people are confident enough in them to accept their forecasts unquestioningly. Consequently, predictions based on com puter models contain sub stantial doses of human judgment. 10 FEDERAL RESERVE BANK OF PHILADELPHIA HOW WELL THEY DID: THE FIRST LOOK age change in the GNP growth rate from one year to the next was just about equal to the average forecast error (2.2 per cent). This raises the possibility that the profes sionals do not do very much better than a simple projection of current trends into the future. One trend projection method based on GNP in the previous three years shows an average error of 2.7 per cent. Every fore caster outperformed the simple trend pro jection method. In some cases, however, the forecasters' leads were distressingly narrow. The first measure of forecasting success that comes to mind is the average size of the difference between what was predicted and what actually happened. Of the 13 GNP forecasters that we put under the micro scope,3 one showed an average error as high as 2.4 per cent of GNP, while another's error was as low as 1.9 per cent on average.4 For all forecasters, the average error was about 2.2 per cent. Whether these numbers are large or small depends, of course, on the standard against which they are compared. In terms of economic goals, overshooting GNP by 2.2 per cent could mean that infla tion turns out to be 4.2 per cent when 2 per cent was anticipated. Alternatively, it might mean that unemployment turns out to be only 3.5 per cent when we were expecting 5 percent. Therefore, in comparison with the usual tolerance levels we put on the econ omy, the typical forecasting error of the '60's could have meant the difference between a "good" year and a "bad" year.5 Measuring forecasters against these toler ances is just one way of testing them. Another is to see how well they do relative to the size of the job. It is obviously much easier to forecast accurately when GNP is growing at a steady rate than when its growth is erratic. During the '60's, the aver THE FORECASTER'S TWO QUESTIONS: WHERE ARE WE? HOW FAR ARE WE GOING? The sword that cuts down the quality of economic forecasts has a double-edged blade. Predictions of next year's GNP are published many months before reliable fig ures are available for the current year.6 So the forecaster must not only predict how much the economy will grow in the new year, but he must also estimate where it stands at the close of the old one. The fore caster can therefore miss the bulls'-eye for two distinct reasons. As a rule, forecasters do not separate their forecasts into estimates of the current year's GNP and predictions for growth next year. Therefore, there is no precise way to sort out one kind of error from the other. But there is a rough way. According to these rough calculations, the average errors in pre dicting the growth in GNP were about onethird lower than average errors in predicting GNP itself. In other words, about one-third of the error in predicting next year's GNP crept in because forecasters added their anticipation for growth to the wrong value of current GNP. The forecasters' greater success in pre- 3 All discussions of group performance are based on whatever data were available from 1959 through 1970. Some forecasters did not start until after 1959, so the early years do not contain data for all forecasters. All official time series used were the latest available at the time of publication and are subject to revisions which might alter some conclusions slightly. 4 Details relating to this and other points appear in the Appendix. 5 An issue that is closely connected to this point is the relationship between forecasting accuracy and the success of national economic policy. For a discussion of this question, see David P. Eastburn, "Forecasting and Policymaking: Some Lessons from Experience," Business Review, Federal Reserve Bank of Philadelphia, April, 1971. 6 The bulk of GNP forecasts are made in November and December. Reasonably reliable figures for GNP are not available until the following July. Minor revisions are frequently made several years later. 11 MAY 1971 BUSINESS REVIEW dieting GNP growth is heartening. For most purposes, we are more concerned with the outlook for growth in GNP than with the outlook for GNP itself. For example, the Administration's predicted $1,065 billion economy is optimistic only because it implies an $88 billion growth in GNP. Ten-sixty-five would be less encouraging if it meant a growth of only $78 billion; this is exactly what it would mean if GNP figures for 1970 were revised upward by $10 billion. BIASES IN FORECASTING An important question that frequently arises is whether there are any systematic biases in economic forecasts. Most people would agree that if predictions are wide of the mark, it is better for the misses to be systematic because then they can be easily corrected if we can count on the "system" remaining unchanged. Even if we cannot, however, we have more confidence that a forecaster "knows what he is doing" if his predictions, though incorrect, show some systematic relation to what actually happens. The difference between random and sys tematic errors shows up very well on what is called a prediction-realization chart. Each dot in the chart (see Chart 1a for an illus tration) stands for one year. The closer a dot is to the top, the higher the prediction for GNP growth in that year; the closer the dot is to the right, the higher actual growth. If a dot lies on the diagonal, it indicates a perfect forecast.7 If the dot for any year falls in the upper triangle, the forecast was too high in that year; if a dot falls in the lower triangle, the forecast was too low. The hypothetical record of forecasts shown in Chart 1a reveals a set of predic7 All points on the diagonal are as close to the top as to the right side of the box. C H A R T I* HOW TO SPOT A “SYSTEM” (a) No Systematic Error Per Cent (b) Downward Bias Per Cent (c) Booms Forecast Too Low, Busts Forecast Too High Per Cent ♦These Charts are merely illustrative, and the labels on the dots are not meant to suggest that they reflect actual values. 12 FEDERAL RESERVE BANK OF PHILADELPHIA tions that has only random error. In contrast, the pattern of dots in Charts 1b and 1c illustrates the two kind of systematic errors frequently made by our 13 forecasters. The hypothetical forecaster whose predictions are reflected in Chart 1b tends to understate the economy's growth potential. Even though Charts 1a and 1b show exactly the same average error, the forecaster whose record is written in Chart 1b has done a better job because his random error is so small. The forecaster associated with Chart 1c tends to understate the intensity of busi ness cycles. As a result, he aims too high when GNP is sluggishly coasting along, and too low when it is soaring upward. Chart 2 shows the actual predictionrealization diagrams of five representatives of CHART 2 SEARCHING FOR A SYSTEM (a) A Consensus Forecaster (b) A Traditional Forecaster from the Large "Herd" (c) A Traditional Forecaster from the Small "Herd" Per Cent Per Cent Per Cent o Actual GNP Growth Actual GNP Growth (d) An Independent Tradi- (e) An Econometric Forecaster tional Forecaster Per Cent Per Cent Actual GNP Growth (f) Trend Projection Per Cent 13 MAY 1971 BUSINESS REVIEW information, we will have to put the 13 fore casters into smaller sub-groups.1* Much has 0 been said about the "herd" instinct among forecasters— the tendency to make predic tions that are similar to one another." One way to group forecasters, therefore, is on the basis of the "herds" to which they belong. Among the 13 forecasters on our range, there are two fairly distinct "herds."1 The 2 larger of the two includes the two consensus forecasters and four "traditional" forecasters (see box). Although these four did not actu ally poll other economists, they apparently used techniques that gave results very much like the consensus predictions. The smaller herd consists of three traditional forecast ers. The remaining four forecasters seem to be "loners" when compared to the other forecasters in the batch: two of them were "traditional" forecasters, and two, "econo metric" forecasters. Not surprisingly, forecasters that fall into the same groups tend to score the same average misses. The six forecasters in the larger goup, for example, tend to do poorly. Four of the six score in the bottom half of the ratings for forecasting GNP, and five of them rank near the bottom in terms of fore casting the growth in GNP. All six of them are in the very lowest rankings when syste matic errors are removed; this is true both of GNP and GNP growth forecasts. Members of the smaller "herd" all have scores that cluster in the medium-to-high rankings. As a group they do decidedly bet ter than the larger pack. When the final scores are in, the winners tend to be drawn from the independents. But the leaders are so bunched that no sin our 13 forecasters8 and for predictions based on trend projection. The diagrams reveal the almost universal tendency to understate the growth rate in GNP. The forecasters' ten dency to understate rapid growth rates and overstate slow growth rates was not nearly so widespread and did not show up very clearly before 1964. (Dots representing 1964 and later years are darker to highlight this point.) The tendency towards downward bias was greater for predictions of GNP (not shown) than for the growth rates. The reason for this is fairly simple. There are two wedges driven between forecasted GNP and actual GNP: the error in estimating current year's GNP and the error in predicting growth next year. The former wedge is removed when growth rates are predicted. Since preliminary GNP figures tend to be too low, forecasters usu ally add their underpredicted changes to an underestimated current GNP. This obviously compounds the bias. The presence of systematic error in the forecasts is beyond doubt. But how im portant has it been? If we remove systematic error from the predictions,9 the average miss falls by roughly one-third, to about 1.5 per cent for forecasts of GNP and 1.1 percentage points for forecasts of GNP growth. Of course, these statistics merely reflect the amount of bias that existed in the '60's. There is no guarantee that the amount or kind of bias will remain unchanged in the future. THE BEST AND THE WORST Investigating forecasters as a group will take us just so far. To squeeze out more 1 0 Not all forecasters' records extend all the way back to 1959. To provide a fair basis for comparison, we compare only the post-'63 records. MThis tendency need not mean that forecasters copy each other's predictions. They may simply use similar methods and similar inputs. 1 2 This grouping is based on correlation analysis. A table of correlation coefficients among forecasters may be found in the Appendix. 8 References in the diagrams to "herds" and inde pendent forecasters will become clearer in the next section. 9 Systematic error was removed through regression analysis. Forecasted values were regressed on actual values. The estimated coefficients were then used to linearly transform the published forecasts and the cal culated forecasts of change. 14 FEDERAL RESERVE BANK OF PHILADELPHIA forecasters for periods no longer than 12 years. But mirage or not, the signs do pro vide suggestions for interpreting forecasts for 1971. One suggestion that flows from the twelveyear experience is that "consensus" predic tions are not particularly good indicators of the future. We would do well, therefore, to stick with the nonconsensus forecasters. But which ones? The answer is, of course, "the best." Un fortunately, a second conclusion of our his tory lesson is that no single forecaster and no single technique— econometric or tradi tional— stood out as being clearly "the best." Regardless of the criterion, the leading fore caster never did very much better than the runner-up, and econometric forecasters could be found at both the top and bottom of the standings. With no clear leader, it will be best to look at the 1971 predictions of more than one forecaster. When we look at predictions of GNP, an econometric forecaster had the smallest total errors and a traditional forecaster had the smallest errors after forecasting biases were removed. The 1971 GNP predictions of these two forecasters differ by about $16 bil lion. This large discrepancy can be almost completely reconciled if we remember an other of our history lessons. Forecasters have followed a pattern in the kinds of errors they make; most particularly, they have tended to aim too low. Adjust for these biases, and the gap between the two predictions falls to $4 billion. One prediction is $1,057 billion and the other, $1,061 billion. This close agree ment is remarkable if unusual. Perhaps more remarkable is the fact that the average 1971 GNP predictions of the top four forecasters1 3 —a "consensus of the cream"— is $1,060 billion. gle forecaster dominated the field during the years we have monitored. If we leave sys tematic errors in, the best ratings for both GNP and GNP growth were earned by one of the econometric forecasters, but very close behind was an independent traditional forecaster. If we remove systematic error, the successful econometric forecaster again leads in a close race for predicting growth; the runner-up is a member of the small ''herd." The award for the smallest random errors in predicting GNP goes to an inde pendent traditional forecaster. Two members of the small "herd" come in right behind, tied for second place. The closeness of all the scores suggests that if the race were run again over some other time period, the rank ings might be different. The fact that one of the econometric fore casters won three of the four first-place awards raises the possibility that this new technique is a major technical advance. Un fortunately, there is not enough evidence to come to a strong conclusion about this ques tion. First, the other econometric forecaster scored lower in the rankings. Second, the quality of any prediction depends to a very large extent on the element of human judg ment. There is no way of knowing whether the successful econometric forecaster did so well because of the mathematical model or simply because the human part of the manmath team had superior forecasting ability. The elusive question is whether this fore caster could predict as well without the model as he did with it. SUMMING UP: LESSONS FOR 1971 PREDICTIONS There are no foolproof divining rods that point to good economic forecasts. Neverthe less, some hazy signs that may help guide policymakers through the wilderness have begun to emerge. The signs may, of course, turn out to be nothing but mirages. After all, we have taken only a limited look at a few 1 3 This "consensus of the cream" prediction is the average forecast of the leaders and runners-up in the races for lowest total error and lowest random errors. Each prediction was adjusted for the forecaster's bias as calculated over the period 1959 through 1970. 15 MAY 1971 BUSINESS REVIEW For most purposes, predictions of GNP growth are more important than predictions of GNP itself. The econometric forecaster that ranked so high at forecasting GNP ex celled also at predicting growth, whether measured by total errors or just random errors. His forecast for growth in 1971 is 7V2 per cent after adjusting for bias. The "consensus of the cream" forecast is a little higher: it is 8 per cent. In short, if the traditional biases in fore casted GNP continue into 1971, actual levels are likely to be higher than the so-called "standard" forecast of $1,045 or $1,050 bil lion and while these biases are not great enough to give unqualified support to the official prediction of $1,065 billion, they do suggest that the Administration GNP fore cast may not be so improbable as many people believe. Things are not so bright when we look at the Administration's GNP growth forecast, however. Traditional biases suggest that the "standard" forecast of 7 per cent growth rate may not be too far off and that the Administration's prediction of 9 per cent1 is too high by at least one per 4 centage point. But we have almost forgotten the most important lesson of all. Economic forecasts are far from perfect, and all of them should be taken with a large grain of salt— including the ones we came up with here. 1 4 $1,065 billion is about 9 per cent higher than the current estimate for GNP in 1970. ■ APPENDIX This Appendix is presented for the reader who has an interest in the details of the study. It contains descriptions of the tech niques used in making calculations men tioned in the text, the various scores and rankings of each forecaster, and the matrix of coefficients of correlation among the fore casters' predictions. GNP Growth Rate. As above, except growth rates replaced levels. HOW FORECASTS FOR GNP GROWTH WERE CALCULATED While final GNP figures are not available for at least six months after year's end, pre liminary estimates are published very early in the new year. Even though these figures are published after the prediction season, they probably represent a fair picture of where economists think the economy is on the eve of the new year. By forecast season, the official estimate of the first nine months' GNP is known. Moreover, scattered informa tion from a variety of sources allows a rough idea of production during the last three months. Consequently, the difference be tween the first preliminary GNP figures for this year and predicted GNP for next year is a crude estimate of forecasted growth. Any revisions in GNP figures are considered as errors in estimating current GNP. HOW TREND PROJECTION FORECASTS WERE CALCULATED GNP. A regression was run for the ten years prior to each year 1959 through 1970. The dependent variable was current GNP; the independent variables were GNP lagged one, two, and three years. The coefficients obtained from the appropriate regression were then applied to the three years prior to the year for which the forecast was to be made to obtain a forecast based on trend. Since final GNP figures are not available in December, one-year lagged data was always based on the first preliminary GNP data. 16 FEDERAL RESERVE BANK OF PHILADELPHIA TABLE 1 E R R O R S IN F O R E C A S T S , A L L Y E A R S Forecaster Classification 1 2 3 4 5 Consensus ” Traditional ” " 6 7 8 9 10 11 12 13 ” ” ” ” ” ” Econometric ” Average error in GNP forecasts as percentage of GNP 2.3 2.4 2.3 2.2 2.2 1.7 1.7 1.5 1.6 1.6 2.3 2.1 2.2 2.3 2.4 2.0 2.1 1.9 Years covered Average error in growth rate forecasts 1.7 1.3 1.5 1.5 1.5 1.3 1.4 0.9 '59-70 '59-70 '61-70 '59-70 ’61,'62, '64-70 '64-70 '59-70 '61-70 '62-70 '59-70 '64-70 '59-70 '63-70 TABLE 2 E R R O R S IN F O R E C A S T S . 1 9 6 4 - 1 9 7 0 Forecaster 1 2 3 4 5 6 7 8 9 10 11 12 13 Classification Consensus Traditional ’’ ” " ” 1 1 Average error in GNP In herd forecasts with as percentage forecasters. . . of GNP 2 ,3 ,4 ,5 ,6 1 ,3 ,4 ,5 ,6 1 ,2 ,4 ,5 ,6 1 ,2 ,3 ,5 ,6 1 ,2 ,3 ,4 ,6 1 ,2 ,3 ,4 ,5 8,9 7,9 7,8 Loner 2.6 2.7 2.8 2.4 2.6 2.3 2.4 2.3 2.5 2.8 2.1 2.3 2.0 M Econometric ” n 11 17 Rank Average error in growth forecasts Rank 9 11 12 6 9 3 6 3 8 12 2 3 1 2.0 1.9 1.8 1.4 1.8 1.7 1.4 1.5 1.6 1.7 1.3 1.7 1.1 13 12 10 3 10 7 3 5 6 7 2 7 1 BUSINESS REVIEW MAY 1971 TABLE 3 R A N D O M E R R O R S IN F O R E C A S T S , A L L Y E A R S Forecaster 1 2 3 4 5 6 7 8 9 10 11 12 13 Average error in GNP forecasts as percentage of GNP 1.5 1.4 1.4 1.4 1.4 1.2 1.1 1.3 1.4 1.2 1.2 1.2 1.0 1.2 1.2 1.4 0.9 Years covered Average error in growth rate forecasts 1.2 1.2 0.9 0.9 0.8 0.8 0.9 1.0 0.8 ’59-70 '59-70 ’61-70 ’59-70 '61/62, ’64-70 ’64-70 '59-70 ’61-70 ’62-70 ’59-70 '64-70 ’59-70 ’63-70 TABLE 4 R A N D O M E R R O R S IN F O R E C A S T S , 1 9 6 4 - 1 9 7 0 Forecaster Average error in GNP forecasts as percentage of GNP 1 2 3 4 5 6 7 8 9 10 11 12 13 1.6 1.3 1.3 1.2 1.3 1.4 1.1 1.0 1.0 0.9 1.3 1.4 1.1 Rank Average error in GNP growth forecasts Rank 13 7 7 6 7 11 4 2 2 1 7 11 4 1.1 1.2 1.2 1.2 1.2 1.2 0.9 1.0 0.7 0.8 1.2 1.2 0.6 6 7 7 7 7 7 4 5 2 3 7 7 1 18 FEDERAL RESERVE BANK OF PHILADELPHIA TABLE 5 M ATRIX O F C O R R E L A T IO N C O E F F IC IE N T S A M O N G P R E D I C T E D G R O W T H R A T E S F R O M 1 964 T H R O U G H 1 9 70 FOR TH IR TEEN FORECASTERS Forecaster 1 2 3 4 5 6 7 8 9 10 11 12 1 1.0 2 0.8 1.0 3 0.6 0.9 1.0 4 0.7 0.9 1.0 1.0 5 0.8 1.0 0.9 1.0 1.0 7 0.2 0.4 0.4 0.4 0.4 0.3 1.0 6 0.9 0.9 0.8 0.8 0.9 1.0 — — 19 8 0.4 0.6 0.5 0.5 0.5 0.6 0.8 1.0 9 0.2 0.3 0.3 0.3 0.3 0.3 0.8 0.9 1.0 10 - 0 .2 0.3 0.4 0.4 0.2 0.1 0.4 0.4 0.3 1.0 11 - 0 .4 - 0 .2 - 0 .3 - 0 .3 - 0 .3 - 0 .3 0.6 0.4 0.4 0.4 1.0 12 0.2 0.4 0.4 0.4 0.4 0.6 0.1 0.5 0.4 0.1 - 0 .4 1.0 13 - 0 .3 - 0 .2 0.1 0.0 - 0 .1 - 0 .1 0.5 0.5 0.7 0.2 0.2 0.5 FOR THE RECORD 2 YEARS AGO YEAR AGO 1971 Third Federal Reserve District SU M M A R Y 2 YEARS AGO YEAR AGO United States Per cent change • • • Per cent change March 1971 from year ago mo. ago 3 mos. 1971 from year ago March 1971 from mo. ago year ago Manufacturing Employ ment 3 mos. 1971 from year ago LO C A L CH A N G ES Standard Metropolitan Statistical Areas* + + + 1 9 7 2 14 2 0 - 8 - 7 - 2 -2 0 + 4 Check Payments** Per cent change March 1971 from mo. ago mo. ago + 8 Per cent change March 1971 from year ago 3 Total Deposits*** Per cent change March 1971 from Payrolls - 5 - year ago mo. ago year ago -1 4 -1 4 + 14 + 23 + 4 + 1 6 1 1 0 CONSTRUCTION** ........ + 102 COAL PRODUCTION . . . . + 3 Banking Per cent change March 1971 from MANUFACTURING Electric power consumed + Man-hours, total* . . . Employment, total . . . . - MAR. 1971 + 33 + 5 + 30 mo. ago year ago 4 Wilmington .. - i - 4 - Trenton ........ + 6 + 12 - 1 + 11 BANKING (All member banks) Deposits ...................... Loans .......................... Investments ................ U.S. Govt, securities . Other .......................... Check payments**'*. . . . 0 - 3 0 + 3 + 32 + 38 0 +28 Altoona ........ 0 - 4 0 + 1 + 4 + 26 + 2 + 12 Harrisburg . . . + 28 + 5 0 - 3 + 2 - 2 + 1 + 9 + 4 + 12 + 1 + 8 + 19 + 1 + 20 Johnstown .. . + 1 + + + + + - 3 2 1 1 1 It + 17 + 10 + 25 + 12 + 33 + 10t + 14 + 10 + 23 + 11 + 32 - 2t + + + + + - 5 1 2 1 3 1 + 19 + 7 + 24 + 21 + 26 + 16 + 15 + 6 + 23 + 20 + 26 + 15 ot + 5$ + 6t •Production workers only ••Value of contracts •••Adjusted for seasonal variation 4 Lancaster . . . . - - 7 - 1 2 - 5 + 6 0 + 86 + 12 + 2 + 17 + 14 + 4 + 17 + 14 + 16 + 7 Lehigh Valley . + 1 - 6 + 2 - 3 + 4 Philadelphia .. - - 8 0 - 1 - 1 2 0 0 + 3 + 5 + 3 + 5 +15 SMSA’s ^Philadelphia Reading ........ - 3 - 6 - 2 - 5 + 13 + 25 + 1 Scranton .. .. 0 - 8 + 1 - 2 - 3 + 2 + 4 Wilkes-Barre . - PRICES Consumer .................... - + 3 1 - 2 0 + 4 + 6 + 12 + 3 + 8 York .............. 0 - 5 0 0 + 4 + 4 + 2 -4 0 •Not restricted to corporate limits of cities but covers areas of one or more counties. ••All commercial banks. Adjusted for seasonal variation. •'••Member banks only. Last Wednesday of the month.
@FedFRASER