The vast majority of business decisions involve some degree of uncertainty and managers seldom know exactly what the outcomes of their choices will be. One approach to reducing the uncertainty associated with decision making is to devote resources to forecasting. Forecasting involves predicting future economic conditions and assessing their effect on the operations of the firm.
Frequently, the objective of forecasting is to predict demand. In some cases, managers are interested in the total demand for a product. For example, the decision by an office products firm to enter the home computer market may be determined by estimates of industry sales growth. In other circumstances, the projection may focus on the firm’s probable market share. If a forecast suggests that sales growth by existing firms will make successful entry unlikely, the company may decide to look for other areas in which to expand.
Forecasts can also provide information on the proper product mix. For an automobile manufacturer such as Maruti Udyog, managers must determine the number of Esteems versus Zens to be produced. In the short run, this decision is largely constrained by the firm’s existing production facilities for producing each kind of car. However, over a longer period, managers can build or modify production facilities. But such choices must be made long before the vehicles begin coming off the assembly line. Accurate forecasts can reduce the uncertainty caused by this long lead time. For example, if the price of petrol is expected to increase, the relative demand for Zens or compact cars is also likely to increase.
Forecasting is an important management activity. Major decisions in large businesses are almost always based on forecasts of some type. In some cases, the forecast may be little more than an intuitive assessment of the future by those involved in the decision. In other circumstances, the forecast may have required thousands of work hours and lakhs of rupees. It may have been generated by the firm’s own economists, provided by consultants specializing in forecasting, or be based on information provided by government agencies. Forecasting requires the development of a good set of data on which to base the analysis. A forecast cannot be better than the data from which it is derived. Three important sources of data used in forecasting are expert opinion, surveys, and market experiments.
The collective judgment of knowledgeable persons can be an important source of information. In fact, some forecasts are made almost entirely on the basis of the personal insights of key decision makers. This process may involve managers conferring to develop projections based on their assessment of the economic conditions facing the firm. In other circumstances, the company’s sales personnel may be asked to evaluate future prospects. In still other cases, consultants may be employed to develop forecasts based on their knowledge of the industry. Although predictions by experts are not always the product of “hard data,” their usefulness should not be underestimated. Indeed, the insights of those closely connected with an industry can be of great value in forecasting.
Methods exist for enhancing the value of information elicited from experts. One of the most useful is the Delphi technique. Its use can be illustrated by a simple example. Suppose that a panel of six outside experts is asked to forecast a firm’s sales for the next year. Working independently, two panel members forecast an 8 percent increase, three members predict a 5 percent increase, and one person predicts no increase in sales. Based on the responses of the other individuals, each expert is then asked to make a revised sales forecast. Some of those expecting rapid sales growth may, based on the judgments of their peers, present less optimistic forecasts in the second iteration. Conversely, some of those predicting slow growth may adjust their responses upward. However, there may also be some panel members who decide that no adjustment of their initial forecast is warranted.
Assume that a second set of predictions by the panel includes one estimate of a 2 percent sales increase, one of 5 percent, two of 6 percent, and two of 7 percent. The experts again are shown each other’s responses and asked to consider their forecasts further. This process continues until a consensus is reached or until further iterations generate little or no change in sales estimates.
The value of the Delphi technique is that it aids individual panel members in assessing their forecasts. Implicitly, they are forced to consider why their judgment differs from that of other experts. Ideally, this evaluation process should generate more precise forecasts with each iteration.
One problem with the Delphi method can be its expense. The usefulness of expert opinion depends on the skill and insight of the experts employed to make predictions. Frequently, the most knowledgeable people in an industry are in a position to command large fees for their work as consultants or they may be employed by the firm, but have other important responsibilities, which mean that there can be a significant opportunity cost in involving them in the planning process. Another potential problem is that those who consider themselves experts may be unwilling to be influenced by the predictions of others on the panel. As a result, there may be few changes in subsequent rounds of forecasts.
Surveys of managerial plans can be an important source of data for forecasting. The rationale for conducting such surveys is that plans generally form the basis for future actions. For example, capital expenditure budgets for large corporations are usually planned well in advance. Thus, a survey of investment plans by such corporations should provide a reasonably accurate forecast of future demand for capital goods.
Several private and government organizations conduct periodic surveys. The annual National Council of Applied Economic Research (NCAER) survey of Market Information of Households is well recognized. Many private organizations like ORG-MARG and TNS-MODE conduct surveys relating to consumer demand across certain geographical areas.
If data from existing sources do not meet its specific needs, a firm may conduct its own survey. Perhaps the most common example involves companies that are considering a new product or making a substantial change in an existing product. But with new or modified products, there are no data on which to base a forecast. One possibility is to survey households regarding their anticipated demand for the product. Typically, such surveys attempt to ascertain the demographic characteristics (e.g., age, education, and income) of those who are most likely to buy the product and find how their decisions would be affected by different pricing policies.
Although surveys of consumer demand can provide useful data for forecasting, their value is highly dependent on the skills of their originators. Meaningful surveys require careful attention to each phase of the process. Questions must be precisely worded to avoid ambiguity. The survey sample must be properly selected so that responses will be representative of all customers. Finally, the methods of survey administration should produce a high response rate and avoid biasing the answers of those surveyed. Poorly phrased questions or a nonrandom sample may result in data that are of little value.
Even the most carefully designed surveys do not always predict consumer demand with great accuracy. In some cases, respondents do not have enough information to determine if they would purchase a product. In other situations, those surveyed may be pressed for time and be unwilling to devote much thought to their answers. Sometimes the response may reflect a desire (either conscious or unconscious) to put oneself in a favorable light or to gain approval from those conducting the survey. Because of these limitations, forecasts seldom rely entirely on results of consumer surveys. Rather, these data are considered supplemental sources of information for decision making.
A potential problem with survey data is that survey responses may not translate into actual consumer behavior. That is, consumers do not necessarily do what they say they are going to do. This weakness can be partially overcome by the use of market experiments designed to generate data prior to the full-scale introduction of a product or implementation of a policy.
To set up a market experiment, the firm first selects a test market. This market may consist of several cities; a region of the country, or a sample of consumers taken from a mailing list. Once the market has been selected, the experiment may incorporate a number of features. It may involve evaluating consumer perceptions of a new product in the test market. In other cases, different prices for an existing product might be set in various cities in order to determine demand elasticity. A third possibility would be a test of consumer reaction to a new advertising campaign.
There are several factors that managers should consider in selecting a test market. First, the location should be of manageable size. If the area is too large, it may be expensive and difficult to conduct the experiment and to analyze the data. Second, the residents of the test market should resemble the overall population of India in age, education, and income. If not, the results may not be applicable to other areas. Finally, it should be possible to purchase advertising that is directed only to those who are being tested.
Market experiments have an advantage over surveys in that they reflect actual consumer behavior, but they still have limitations. One problem is the risk involved. In test markets where prices are increased, consumers may switch to products of competitors. Once the experiment has ended and the price reduced to its original level, it may be difficult to regain those customers. Another problem is that the firm cannot control all the factors that affect demand. The results of some market experiments can be influenced by bad weather, changing economic conditions, or the tactics of competitors. Finally, because most experiments are of relatively short duration, consumers may not be completely aware of pricing or advertising changes. Thus their responses may understate the probable impact of those changes.