Corporate Planning The future isn't what it used to be -- Anonymous Human fascination with the future can be traced back to prehistoric times. People have always wanted to predict the future for psychological reasons (to reduce their fear and anxiety about the unknown). Concrete evidence shows that the human desire to foresee the future has been exploited by some for profit, fame, or power. There is a scientific basis to forecasting; but there are also serious limitations on our ability to predict future events and situations. This chapter shows how you can be convinced about someone's alleged abilities to predict the future and how it is necessary to avoid being misled into believing in prophecies. It becomes important to know precisely what can and cannot be predicted, and the advantages as well as the limitations of forecasting. The temple of Delphi is in the middle of the Parnassus mountain range, on the southern part of the Greek mainland. The view from it is awe-inspiring; excellent visibility allows one to see a good part of the Gulf of Corinth and the northern Peloponnesus. The blue of the sky and the sea are perfectly blended with the green mountains to provide a unique breathtaking image, difficult to forget. In such a setting the priests of Delphi operated the first institutionalized forecasting service. To reach the temple required a long and tiring mountain climb, but the hospitable priests gave their guests a warm welcome. Good food and plentiful wine were made available to the exhausted travelers who, while eating and drinking, were encouraged to talk about themselves and about their needs, desires, and expectations. The temple of Delphi became the richest institution in ancient Greece. Its power of prophecy was believed in by everyone, because it was attributed to Apollo, god of the Sun, who could see and therefore foretell the future for all those who could afford the services of his oracle. The oracle, or sibyl, gave her predictions in such a way as to make their invalidation difficult. Those seeking advice were often told what they were expecting to hear. If that was not possible, the prophecies were equivocal: their wording was obscure, they were general or they were impossible to check against reality. EXAMPLES OF DELPHI PROPHECIES Telling what one wants to hear: I do not know if you are a god or a human, but I see more of a god in you. Double meaning or conditional wording: If you cross the river a great army will be destroyed (whose army is left unsaid). Obscure or general wording: If you return to Athens you will gain the benefit of the law (but laws can condemn someone to death too). Cannot be checked against reality: Sophocles is wise, Euripides is wiser; but Socrates is the wisest. Delphi was not the only place selling prophecies. Since the dawn of civilization priests, astrologers, prophets, fortune tellers, and the like have sought to satisfy the human need to predict the future. The temple of Delphi became the best known and most successful of all because of the uncanny shrewdness of the priests who ran it. They were full-time employees, the first professional forecasters, who had an obvious interest in making people believe that their oracle could foretell the future. Their livelihood and prosperity depended on that accomplishment. Judging from the unquestioned acceptance of their prophecies, the richness of their holdings, and their lasting influence (more than 500 years) we must conclude that theirs was a success story, although nobody would doubt today that the predictive power of their prophecies was zero. Are things any different at present? EXAMPLES OF PRESENT-DAY PROPHECIES SIMILAR TO THOSE OF DELPHI Foretelling what we want to hear: The Dow-Jones industrial average will hit the 3,500 mark by the middle of next year. Double meaning or conditional wording: All evidence indicates that a great depression is in the making, and unless immediate remedial action is taken, the results will be catastrophic. Obscure or general wording: To the uninitiated, markets can often seem perverse. But recent weeks have added a bizarre new twist. The faster the U.S. economy expands, the more anxiously Wall Street seems to worry that it will contract. Cannot be checked against reality: Unless pollution is stopped, carbon dioxide will increase the earth's temperature resulting in the melting of the polar ice in 30,000 years. Today forecasting is a multibillion dollar industry. All economic publications devote considerable space to economic forecasts; political writers hold forth on political trends and forthcoming government policies; stockbrokers and financial gurus foretell stock market trends, when to buy, and what stocks to select; and many others prophesy on such diverse topics as when the next ice age will come (or alternatively the melting of the polar ice), how to cope with the inevitable great depression of the 1990s, what to do when a barrel of oil costs $100 in 1995, or how to survive after a nuclear World War III. Unfortunately, many people believe such forecasts, prolonging the myth that somehow prophecies are possible. Even worse, there are people who seriously believe they possess divine powers enabling them to predict the future. Worst of all, reputable publishing houses and widely read business journals publish such forecasts, perpetuating the expectation and the myth that they can be usefully implemented to improve future-oriented decisions. In the next section I shall show that even today superstitious beliefs about the future abound. The empirical evidence proves beyond any reasonable doubt that no one can consistently predict the future more accurately than his or her colleagues. Yet there are eminent forecasters and famous gurus who pretend to be capable of foreseeing the future. STOCK MARKET FORECASTS: THE MODERN VERSION OF DELPHI Stock market investing is an important activity in our day. One can become rich if the right stocks are bought or can lose huge sums through the wrong choices. Today there are about 800 newsletters offering stock-picking advice. Every general financial publication offers daily advice about the stock market and individual stocks. Value Line http://www.valueline.com/ analyzes all listed stocks and evaluates their performance; every major brokerage house employs its own gurus, and even small brokerage firms have at least one forecaster on their payrolls. All this generates a huge volume of stock market forecasts. Millions of people read them -- some paying steep subscription fees to receive stock market newsletters -- and act upon their advice. Obviously, a lot of the issuers of stock market forecasts become rich in the process, and the livelihood of many others depends upon their predictions. Yet the value of all these forecasts is zero. This sounds like Delphi, doesn’t it? Is it possible that mass deception still exists in our day? I believe it does, and I propose to explain why. THE RANDOM WALK THEORY In 1965 Eugene Fama of the University of Chicago (see http://www.ssrn.com/ and http://www.ibbotson.com/research/other/fama/fama.asp) published an influential paper in which he proved, using theoretical reasoning and empirical evidence, that predicting changes in stock market prices (increases or decreases from ''today's'' value), either as a whole or for any individual stock, could not be done any better than by using "today's" price (taking the closing stock prices reported on the financial pages of newspapers) as the forecast. This means it is impossible for investors to profit from so-called accurate forecasts of which stock prices will go up or down. Indeed, nobody can do so without inside knowledge, which cannot be used legally. Then why are there 800 newsletters and daily newspaper reports, and why are there stock market gurus? Obviously, their existence makes no sense from a rational point of view. Both the theoretical and the empirical evidence are indisputable: No one can predict stock market prices more accurately than by using ''today's price. Yet the overwhelming majority of investors (both individuals and institutions) still prefer a stock analyst or a professional manager to take care of their money. By sheer luck, some analysts or managers might do better than a market average for a certain period of time, but all evidence suggests that consistent above-average performance is virtually impossible. Thus, for example, you could hang the Wall Street Journal listing of NYSE stocks on the wall, throw ten darts to choose ten stocks, and your portfolio, on the average, will perform no worse than one selected by an expert. However, few people feel comfortable with such a random selection when investing their life savings. The random walk (changes in stock prices are random, hence unpredictable) is a highly accurate predictive theory. If you read Business Week, The Economist, Forbes, and similar periodicals (see http://ceohomepage.com/bizpubs.htm and http://ceohomepage.com/newspubs.htm), you can find ample evidence of its accurate predictive value. What surprises me is that virtually all executives in the courses I teach are not aware of this evidence, although they also read Business Week, The Economist, Forbes, and the rest. So let me summarize some of the evidence (more on market efficiency at http://www.deanlebaron.com/book/ultimate/chapters/mkt_eff.html). If stock market experts could forecast better than you, I, or anybody selecting stocks at random, professionally managed portfolios and mutual funds would perform better than the market average. That is not true and has rarely been true since such comparisons were started in the late 1920s . Here are some recent statistics concerning average returns on stocks, collected from various 1987 Business Week issues:
The same results can be found in average returns on money invested in bonds. Managed bond funds do not outperform the market average. Here are some recent statistics:
One could say that the above comparisons are not fair because they include all professionally managed funds and all mutual funds. It might be true that some experts do better than the rest. Forbes regularly asks ten experts to recommend their favorite stock. The most recent available results when this chapter was written show that the ten stocks selected by the experts lost an average of 22 percent, against the market's overall loss of 4 percent. The above evidence demonstrates the random walk theory. Random selection of stocks or bonds (which indicates average performance) outperforms selection by expert forecasters. The important question is whether or not John Q. Investor believes this evidence and will act upon it. In the first chapter I recommended being a skeptic, which in this case necessitates a search for additional evidence. Paying careful attention to stock market and bond comparisons when reading financial publications clearly reveals that Delphi-type forecasts abound even in our day. Individuals and companies can save millions in fees by not using them. If the random walk is valid, how do some people get rich by playing the market? Some people also get rich by playing in casinos. Does that mean they can correctly predict the winning numbers? If there are millions of people playing the market, it is inevitable that some will make lucky choices. In 1966 the publishers of Forbes magazine selected twenty stocks by throwing darts at the stock market page of that day's Wall Street Journal. Fifteen years later their randomly selected portfolio had risen 239 percent, while the S&P 500 was up only 35 percent. Luck is possible even in random selections. If someone places a bet on a roulette number, his or her chance of winning is one out of thirty-seven. Since a lot of people play, some will win. Suppose a winner puts all he or she won on another number. He or she will lose 36 out of 37 times, but since there are many people playing, someone will inevitably win twice in a row (the chances are that one person out of 1,369 will). But since a large number of people play, someone can win three times in a row (the chances are that one out of 50, 653 people will do so). Thus, depending on the upper limit permitted on a single number, he or she can become a millionaire. No skills are involved, just plain good luck. Luck plays strange tricks. Robert E. Humphries, a truck driver, won two Pennsylvania lottery jackpots in a period of less than two years, which brought him net annual earnings of more than $250, 000 for life. He spent only $45 a week on the game. Was Mr. Humphries skillful in picking the winning ticket? Although the chance of winning two jackpots in such a short time is more than one in a billion, someone is bound to win, since an extremely large number of people buy lottery tickets. Again, no skills are involved, just plain good luck. Does the fact that stock market predictions are of little value mean that forecasting is useless? The answer is a categorical no. Forecasts are useful in 3 wide range of applications. However, it is important to know why forecasts are needed and when they can be used to gain concrete benefits, or at least to understand the uses, advantages, and limitations of forecasting. WHY ARE FORECASTS NEEDED? Forecasts are made with a purpose and an audience in mind. There are three principal uses of forecasting: 1. Satisfying Curiosity Some forecasts aim to satisfy simple curiosity. Horoscopes, astrology, and fortune-telling fall into this category. So do science fiction and the vast majority of forecasts whose purpose is to describe distant future worlds and what they might be like. Writings about interplanetary travel, intelligent robots, and brand new energy sources fall into this category. For managers, such type of forecasts provide little of direct value. However, if superstitious beliefs can be separated from objective, rational projections, there might be some indirect value, because such forecasts might help see the future in a clearer way. Managers could better anticipate the future by establishing long-term visions and formulating general strategic directions that could help their organizations change in order to cope better with what is ahead. But nothing is certain; no one can be sure whether, and most importantly when, the forecasts will become reality. 2. Improving Decision-Making Some forecasts aim at improving the value of decision-making. Stock market predictions, the start and beginning of the next recession, the number of cars to be sold in the last quarter of next year, and the party that will win the next election are forecasts whose accuracy can benefit or, should they be wrong, damage the person or organization using them as a basis for decisions. In the introduction. I talked about suicides and murders directly related to false expectations (forecasts) about the stock market. Forecasts in this category are not neutral. They can be harmful as well as beneficial, and great care must be taken while using them. It is therefore necessary to know how accurate they are. In addition to being accurate, forecasts must also be reliable. A forecaster might be accurate 90 percent of the time but the remaining 10 percent might produce disastrous results. Knowing the degree of uncertainty associated with the forecasts can be as critical a factor as accuracy itself when the objective is to use the forecasts to improve future-oriented decision-making Many forecasts are as inaccurate and unreliable as those of fortune tellers, but most decisions require forecasts whose accuracy and reliability or uncertainty must be assessed correctly. 3. Generating Consensus The last category of forecasts aims at generating consensus, an important and useful forecasting function little talked about, little understood, and even less exploited. The politicians in ancient Greece used the oracle of Delphi as their main vehicle to achieve consensus. If the god had said something, it was difficult for mortals to argue against it. Obviously no divine power can be called upon today to certify forecasts. Nevertheless, forecasting can play an important role in generating consensus among top management concerning desired ways in which to move the organization. Scenarios and long-range forecasts fall into this category. Each of the three types of forecast described above is made using different procedures, aims at satisfying different needs, and requires different criteria for evaluation. THE SCIENTIFIC BASIS OF FORECASTING Although some animals (like bees and ants) do possess some concept of the future, humans are unique in their ability to comprehend and plan for a wide range of future events. Humans can predict the future by observing regularities (patterns) in certain phenomena (the daily sunrise or the seasons) or causal relationships (cultivating seeds and growing crops, or intercourse and pregnancy). A prerequisite of any form of forecasting, whether judgmental or statistical, is that a pattern or relationship exists concerning the event of interest. If such a pattern or relationship does exist and can be correctly identified, it can subsequently be projected in order to forecast. Consider the forecast of how long it takes to get to work after leaving home. Years of experience lend accuracy to such predictions. Some days, or during certain hours, it takes longer. On days when it is raining heavily the commuting time becomes longer but not as long as it is during snowstorms. when it might take double the usual time. Those patterns (time of daily commuting) or relationships (rain increases commuting times, snow even more) exist in the commuter's head and help him judge approximately how long it will take to get to work on different days of the weeks at different times of the day, or in various weather conditions. Scientific forecasting uses exactly the same principles as those in the commuter's mind, but it does so with more precision and more objectivity. Scientific forecasting uses quantitative data to measure exactly how long it will take to go from home to work. It would require a stopwatch set upon leaving home and stopped on arrival at the office. It would also demand a record of the day of the week, the time of departure, the amount of rainfall, the amount of snowfall, if it is cloudy or sunny, and whatever other factors the commuter himself or an expert might consider important. A statistical (mathematical) model is then employed to identify the important factors influencing daily commuting time and to measure the extent of such influences. Once determined and measured these factors (variables) can be used to forecast the commuting time for each day of the week, according to the specific time of departure and the weather conditions. The more data (daily observations) included, the more reliable the forecast will usually be. All forms of scientific forecasting use precisely such a procedure. First, data are found or collected; second, a statistical model is selected; third, the patterns and relationships involved are identified and precisely measured; fourth, forecasts are made by projecting the patterns and relationships identified and measured; finally, the uncertainty of the forecasts can be estimated (for example, an accident blocking the highway can make commuter time longer than average) and used to reduce the negative consequences of unexpected events. In many areas of the natural sciences or physics, forecasting accuracy is perfect for all practical purposes. Thus, the exact timing of the sunrise tomorrow or a year from now can be predicted to the thousandth of a second. Similarly, one can also predict that water will form if two atoms of hydrogen and one of oxygen are put together. However. such success does not extend to all areas of natural and physical sciences. Prediction in the social sciences, and notably in the economic and business fields, is much more difficult than in the natural sciences. FACTORS AFFECTING FORECAST ACCURACY The accuracy of your forecasts will be determined to a large degree by how much patterns and relationships change, and how much people (including the organization itself and its competitors) can influence future events. Patterns or relationships might change over time. A critical assumption for accurate forecasting is that patterns or relationships, once identified and measured, remain constant. In the commuting examples a new highway or the closing of an existing one for repairs will influence the flow of traffic, causing patterns and relationships to change. Such a change will make forecasting inaccurate until several trips under the new conditions have been made and new estimates drawn. Despite the fact that satellite photos and huge computer models tracking weather patterns are being used at present, weather forecasts are sometimes wrong. The reason is that certain weather patterns do not always follow the same direction, nor do they move at a uniform speed. Both direction and speed can and do change, and new weather patterns develop that modify or cancel out existing ones. At other times, calm might prevail between two storms blowing in opposite directions. Weather forecasters can predict fairly accurately when weather patterns do not change. However, they do not know when or how they will change. Their huge mathematical models, the super-fast computers they use, the multimillion-dollar specialized satellites, and their sophisticated weather stations have not increased the accuracy of weather forecasts much over the last thirty years, simply because changes in weather patterns cannot be predicted. Although weather patterns can be better tracked and more accurately predicted if they do not change, inaccuracies inevitably develop when they do change. In the economic and business fields, patterns and relationships change much more, and much more often, than those of the weather. People can influence future events. In the economic and business environment, predictions can become self-fulfilling or self-defeating prophecies, nullifying the forecasts. Consider, for instance, a forecast of a recession that is going to arrive during an election year. Any party in power that wants to be reelected will take action to avert the predicted recession, thus rendering inaccurate what could otherwise have been a perfect forecast. Similarly, an attractive opportunity for investment might bring big losses if several competitors arrive at the same forecast. The fact that the forecasts themselves can influence future events and in so doing change their course complicates the task of forecasting. It is no longer sufficient to predict accurately what is going to happen -- managers must also forecast what competitors will do in response to such predictions. Because people can change the course of future events, the task of forecasting becomes much more difficult. This is different from the natural world, where humans still cannot change future events, such as the weather. Here are some additional factors affecting forecast accuracy: 1. The Time Horizon of Forecasting The longer the time horizon of the forecasts, the greater the chance that established patterns and relationships will change, invalidating forecasts. Specifically, the more time competitors have to react to predicted events or the predictions themselves, the more able they will be to influence future events for their own benefit. Thus, all else being equal, forecasting accuracy decreases as the time horizon increases. 2. Technological Change The higher the rate of technological change in a given industry, all other things being equal, the greater the chance that established patterns and relationships will change, and the greater the chance that competitors will be able to influence the industry through technological innovation. An excellent example is high-tech industries, where forecasting is almost impossible as firms strive to create the future according to their own conceptions. By bringing out new technologies, they hope to shape the future in desired directions in order to achieve competitive advantage. Thus, forecasting accuracy decreases as the rate of technological change increases. 3. Barriers to Entry The lighter the barriers to entry, all other things being equal, the more inaccurate the forecasting, as new competitors (both domestic and foreign) can drastically change established patterns and relationships in their quest to gain competitive advantage. 4. Dissemination of Information The faster the dissemination of information, all other things being equal, the less useful the value of forecasting, as everyone will have the same information and can arrive at similar predictions. In such a case it becomes impossible to gain advantages from accurate forecasting, as everyone else will also attempt to do so. This means accurate forecasts are not necessarily useful, a point that is not always understood or accepted, although examples abound. The growth in mainframes and microcomputers was correctly predicted, but few gains resulted, as many companies that used such accurate forecasts went bankrupt. 5. Elasticity of Demand The more elastic the demand, all other things being equal, the less accurate the forecasts. Thus, demand for necessities (for example, food items) can be predicted with a higher degree of accuracy than for non-necessities (such as vacationing). Obviously, people must eat and acquire necessities, which are given priority over other purchases in case of income reduction, as during periods of recession. 6. Consumer Versus Industrial Products Forecasts for consumer products, all other things being equal, are more accurate than those for industrial products. Industrial products are sold to a few customers. If only one of those customers is lost, the resulting error can represent a substantial proportion of sales, because of the large quantities, or sales value such customers buy. Those customers are well informed and can receive offers of bargain terms from competitors because of the large quantities or value amounts they buy. WHAT CAN AND CANNOT BE PREDICTED, AND THE IMPLICATIONS Over the last thirty years a considerable amount of empirical evidence and experience with forecasting applications has been accumulated. At present it is much easier to decide what can and cannot be predicted and what can be done to benefit from, as well as avoid the dangers of, forecasting. From the study of past forecasts a general observation becomes clear: The vast majority of forecasters underestimate future uncertainty. This bias was discussed in the last chapter and has to be reemphasized here. Uncertainty is underestimated when expected events do not materialize and when unexpected ones occur. In both cases there is surprise which could have been avoided if decision-makers were willing to recognize that future uncertainty is more extensive than it appears. SHORT-TERM PREDICTIONS In the short term, forecasting can benefit by extrapolating the inertia (momentum) that exists in the economic and business phenomena. Seasonality can also be predicted fairly well. Empirical evidence has shown that seasonality does not change much. Thus, once computed it can be projected, together with the momentum of the series being forecast, with a high degree of accuracy. The momentum in series and their seasonality constitute the two greatest advantages that can be gained by using formal forecasting methods. Such advantages can be of benefit in production planning and scheduling; equipment, personnel and financial planning; and the ordering of raw and other materials. As seasonal fluctuations can be substantial, accurate prediction can greatly improve short-term scheduling and planning decisions. The larger the number of customers or items involved, the smaller the effect of random forces and the higher the reliability of forecasting. Thus, firms selling to consumers not only can forecast more accurately but can know that the uncertainty of their forecasts is less than that of firms selling to industrial customers. Estimating uncertainty can be used to determine safety stocks (for finished products and materials), slack in personnel and equipment, and financial reserves, so that possible errors in forecasting can be confronted with a minimum of surprise and unpleasant consequences. Short-term forecasting and the estimation of uncertainty are technically feasible and can be employed on a routine basis to provide improved customer satisfaction, better production and service scheduling, and so forth. If an organization does not already use a statistical, computerized system to make short-term forecasts and estimates of uncertainty, my advice would be to do so as soon as possible. Overwhelming empirical evidence shows concrete benefits from using simple statistical methods instead of using judgment to make the forecasts and to estimate uncertainty. Human forecasters cost a company a great deal more and are not necessarily more accurate. As a matter of tact, the vast majority of empirical comparisons clearly show the superiority of simple statistical methods over all other alternatives, including judgmental forecasts. Although few things can happen in the short term to alter established patterns, some changes are occasionally possible, introducing an additional element of uncertainty. For instance, unexpected events (a fire, a major machine breakdown) or special events (a big snowstorm) can take place, or competitors can initiate special actions (advertising campaigns, price decreases, or the like). Such unexpected events or actions can change established patterns, thus invalidating the forecasts and introducing additional uncertainty. MEDIUM-TERM PREDICTIONS In the medium term, forecasting is relatively easy when patterns and relationships do not change. However, as the time horizon of forecasting increases, so does the chance of a change in established patterns and relationships. Economic cycles, for one thing, can and do change established patterns and relationships. Unfortunately, however, we have not yet been able to predict accurately the timing and depth of recessions or the start and strength of booms. This makes medium-term forecasting hazardous, as recessions and booms can start any time during a planning horizon of up to two years (the usual length of the medium term). In addition, the uncertainty in forecasting becomes greater and less easy to measure or deal with, because the differences between forecasts and actual results can be substantial, especially in cyclical industries. Medium-term forecasts are needed mostly for budgeting purposes. They require estimates of sales, prices, and costs for the entire company, as well as for divisions, geographical areas, product lines and so forth, and predictions of economic and industry variables. In the case of a business cycle (a recession or a boom), all variables being predicted will be influenced in the same direction and by similar amounts, thus causing large errors that might necessitate the closing down of factories, the firing of workers, and other unpleasant belt-tightening measures. The deeper the recession, the worse the forecasting errors and the greater the unpleasant surprises and negative consequences. During a boom the opposite type of error normally occurs, giving rise to underestimated demand, personnel needs, and the like. Medium-term forecasts can therefore become inaccurate. Worse still, their uncertainty is difficult to measure. Although there are forecasting services and newsletters claiming to be able to predict recessions or booms, empirical evidence shows beyond any reasonable doubt that they have not been successful up to now. This means that our inability to forecast recessions and booms must be accepted and taken into account during budgeting processes and when formulating overall strategies. On the other hand, the end of a recession (once it has started) is easier to predict. Recessions last about a year, and their length does not fluctuate widely around the average. Not all companies are equally affected by cycles. In general, manufacturing firms are more affected than service firms; firms producing or servicing luxury (elastic) goods are more affected than those producing or servicing necessities (inelastic products and services); industrial firms are more affected than consumer firms; and companies in industries where strong competition exists are affected more than those in less competitive ones. In dealing with business cycles it is important to remember that recessions or booms do not last forever. Past history clearly shows that cycles are temporary and will continue to be so unless some fundamental change in the business and economic environment takes place, which is unlikely at present. Thus, when a company is in a recession, managers must plan for the recovery. If there is a lasting boom, they must be concerned about the coming recession. In forecasting and planning there is one thing that is sure: After a long boom a recession is inevitable. The only thing not known is when it will start and how deep it will be. Thus, contingency planning to face the coming recession becomes necessary. The same is true during periods of recession. A recovery is certain. The only question is when it will start and how strong it will be. Obviously, there is always the possibility that a recession might turn into a depression or even that it might never end. Although such a possibility exists, it is highly unlikely and cannot be seriously considered; a firm trying to provide for it will have to be too conservative in its plans and strategy and will be overtaken by more aggressive competitors. Although we know that a car can hit us when we are crossing a street, no one can seriously consider never walking because of the possibility of being killed by a passing car. Because recessions and booms cannot be predicted, it becomes necessary, over the medium term, to monitor for possible recessions or booms. This is the second best alternative to forecasting. It is like having a tracking system of radars looking for a possible enemy attack. It cannot tell us when the attack will be launched, but it can warn us once it is on its way. Although monitoring is not forecasting, it helps managers not to be taken completely by surprise by the arrival of a recession or boom. Another way of anticipating recessions is by looking for imbalances in one's own industry, in the economy, or in the international financial system. The bigger and the more widespread such imbalances, the greater the chance of a correction, which usually takes the form of a recession or a boom. In practical terms, it makes little sense to attempt to forecast recessions or booms. Recessions or booms that are not forecast occur, and others that are predicted do not materialize. Therefore, spending money or resources to predict recessions or booms adds little or no value to future decision-making. It is best to accept that such a task is not possible and to plan budgets by extrapolating established trends and relationships. The company should be capable of adjusting its plans as soon as monitoring has confirmed a recession or boom. This is where contingency planning can be of great help. As recessions or booms are certain to arrive, managers can be prepared to face them by having drawn up detailed contingency plans. LONG-TERM PREDICTIONS Long-term forecasts are needed mostly for capital expansion plans, selecting R&D projects, launching new products, and formulating long-term goals and strategies. The critical element in long-term forecasting is the prevailing trends. The challenge is to determine when and how such trends might change and how societal and consumer attitudes will differ in the future. The chances that there will be changes in long-term trends caused by new products, new services, new competitive structures, new forms of organization, and other novelties are great, making the task of forecasting difficult but also of critical importance. I prefer to divide the long-term into three types: emerging, distant, and faraway. In the emerging long-term (two to five years), most changes requiring consideration have already started. It therefore becomes a question of figuring out their effects on the given organization and what can be done to deal with such changes. A common mistake, repeated over the span of the emerging long term, is to ignore technological and other changes until a crisis point has been reached, in which case organizations often overreact (for instance, AM International or Western Union). An equally common mistake is to be dazzled by the technological wonders of new inventions and to rush into introducing them (the picturephone, robotics). Many new technologies turn out to be uneconomical or to require more time before they can be smoothly implemented (smart credit cards http://www.smart-card.com/, video text). Besides, new technologies are initially expensive, making it uneconomical and unwise to rush into adopting them (picture phones, smart credit cards, and so on). When moving to the distant and faraway long-term, the accuracy of specific forecasts decreases drastically, as many things can happen to change established patterns and relationships. The purpose of forecasting in such cases is to provide general directions as to where the world economy, or a particular industry, is heading, and to identify the major opportunities as well as the dangers ahead. The foremost challenge is to predict technological innovations and how they will affect the organization. New technologies can drastically change established demand, societal attitudes, costs, distribution channels, and the competitive structure of an industry. The major purpose of such long-term forecasting is to help the organization form a consensus about the future and start considering ways of adapting new technologies once they become economically profitable. Distant and faraway long-term forecasts cannot be specific and will always be highly uncertain. Thus, their value lies not in improving decision-making but in helping generate organizational consensus. THE EVALUATION OF FORECASTS A manager is constantly bombarded with forecasts. How does he or she distinguish the Delphi type from forecasts that are accurate and useful? It is not easy. Present-day forecasters have adapted their selling approach to modern times. Computer printouts and fancy mathematical models are often used to legitimize the forecasts and provide them with an aura of scientific objectivity. Here are some guidelines for evaluating the accuracy and usefulness of forecasts: 1. Use of Benchmarks A manager can compare the forecasts given to him with those of simple benchmarks. For instance, are weather forecasts more accurate than predicting that tomorrow's weather will be the same as today’s? Is an exchange rate forecasting model being sold for $30, 000 more accurate than using today's actual exchange rate for forecasting next year's? When using simple benchmarks, we often find no difference, or not enough difference to justify the cost of generating or buying the forecasts. Empirical evidence has shown that in a great many cases simple, naive forecasting models outperform complex or sophisticated methods and judgmental forecasters. 2. Track Record of Past Forecasts Sales of forecasting services or models are often promoted by emphasizing how accurately they have predicted or performed in the past. That does not mean much. Past forecasting successes do not guarantee future dependability. Consider for instance, a mutual fund that has increased its assets 20 percent. If the S&P 500 index has increased 25 percent (in which case the S&P becomes the benchmark), the fund's record is not impressive. Even if the fund outperforms the S&P 500 for many years, that is not a sure sign that it will do so in the future. Considerable empirical evidence shows that no forecaster, forecasting service, or forecasting model consistently outperforms the rest. The only accuracy comparisons that count must be made against the future, when conditions will be different from those of the past. Unless a manager can have a look at such future-oriented comparisons (which are impossible to make), he should think carefully before buying forecasting services, newsletters, or models simply because of their superior track record. Other reasons must be offered to prove their worth. Empirical evidence has clearly shown that averaging the predictions of several forecasters or models provides more accurate forecasts than the individual forecasters or the specific models. 3. Assumptions Used Scientific forecasting can accurately predict (and estimate uncertainty) when established patterns and relationships do not change. Any other forecasts are not possible to make except by using analogies (for example, sales of a new product will follow the same pattern as those of a similar existing product), in which case the analogy must be made clear, or by making subjective inferences as to how patterns and relationships might change. Making analogies or subjective inferences, however, requires several assumptions, which must be made explicit, as the accuracy of forecasts relates directly to the validity of those assumptions. A manager can then judge the appropriateness of the forecasts himself by evaluating the validity of the analogy or the correctness of the assumptions. 4. The ''Don'ts'' of Forecasting Don't elieve any forecasts based on secret formulas or complex computerized models that cannot be explained because they are beyond a manager's comprehension. No such formulas or models have been proved to be of value in the economic and business fields. Don't believe anyone who pretends to possess prophetic powers. Such people are charlatans or, worse, fools who should be treated as such. There is very strong evidence to show that complex or sophisticated forecasting models do not outperform simple ones, and that people do not forecast more accurately than simple statistical methods. Thus, if the accuracy of a forecast is low, don’ t attempt to improve it by introducing more complex or sophisticated approaches or by substituting people for statistical models. It is not likely that they will help increase your forecasting accuracy . THE PARADOXES The future can be predicted only by extrapolating from the past, yet it is fairly certain that the future will be different from the past. Through forecasting we would expect to reduce future uncertainty, yet as we consider the future more carefully, we realize that unexpected events are possible, thus increasing uncertainty. Although forecasts can and will be inaccurate and the future will always be uncertain, no planning or strategy is possible without forecasting and without estimating uncertainty. CONCLUSIONS Forecasting the short-term can be done accurately on a mechanical basis in most business situations (although accuracy will vary from one case to another). Uncertainty can be estimated reliably in the short-run. As the time horizon increases, so does the difficulty of arriving at accurate forecasts and reliable estimates of uncertainty, as well as the challenge of planning rationally and formulating successful strategies. Can a manager plan and develop effective strategies without accurate forecasts and reliable estimates of uncertainty? There is not much choice. Planning and strategy formulation must be done in spite of inaccurate forecasts and high levels of uncertainty. Success will depend on taking educated risks, being able to forecast changes in established patterns and relationships more accurately than competitors, and taking effective action to anticipate such changes. EVENTS OR SITUATIONS THAT CAN BE FORECAST AND THE NEED FOR COMBINING Short Term Inertia/momentum Seasonality Normal uncertainty Medium Term Average growth rates (assuming no recession or boom) Expected costs and revenues (assuming average growth rates and average changes in costs and prices) Long Term Established trends Existing relationships Averaging (Combining) of Forecasts When in doubt about which forecaster or forecasting model to use, average the predictions of several models or forecasters and employ the average to forecast; empirical evidence has shown beyond doubt that averaging improves forecasting accuracy and reduces uncertainty FORECASTING CHALLENGES Short Term Special events and actions and their influence on the future Dealing with the uncertainty that can be caused by such special events and actions Medium Term Recessions and booms Dealing with the uncertainty that can be caused by recessions and booms Long Term Changes from established patterns and relationships, in particular those produced by important technological innovations Developing appropriate strategies to anticipate and deal with the uncertainty caused by such changes Source MAKRIDAKIS, S. G. (1990). Forecasting, planning, and strategy for the 21st century. NY: The Free Press. (Ch. 3. Predicting the future: Myths and reality, pp. 49-68). Homepage of Professor Makridakis at http://www.insead.edu/facultyresearch/faculty/profiles/smakridakis/ |
