Forecasting Methods for Management

Managers frequently need to make decisions about the future of the organization. Forecasting is the science of estimating or predicting future trends to support managers in this process. Forecasting methods can be used to provide information to support decisions about many aspects of the business including buying, selling, production, and hiring. Many statistical techniques are available for use in forecasting. However, each is not equally applicable in every situation. In addition to quantitative methods for forecasting, there are also subjective or qualitative forecasting methods that are used by many managers. Experienced and insightful managers can take advantage of years of experience to extrapolate trends in ways that are still not possible through the use of quantitative techniques alone. As a result, quantitative and qualitative analyses are inseparable for most forecasts.

Every day, managers are faced with decisions that need to be made. Some of these are simple such as the routine reordering of supplies or approval of timesheets. Others are more complex, such as determining how to rate someone in an annual review or determining who should be included in the company's layoffs. Another category of complex decision making that managers face is forecasting. This is the science of estimating or predicting future trends. Forecasting is used to support managers in making decisions about many aspects of the business including buying, selling, production, and hiring. For example, managers need to be able to predict the demand for a product or service over a given time period. This will allow them make a number of other decisions. If there will be an increased demand for the organization's product, management can feel confident that they can meet their financial obligations for that time period. However, they may also need to hire additional workers, lease additional facilities, and acquire or store additional raw materials or components to meet the increased demand. Further, a reasonable forecast about demand can also enable the organization to make better strategic decisions about where to take the business line in the future, whether or not to invest in an additional product line, and so forth. On the other hand, if the organization knows that there will be a decreased demand for their products or services for the foreseeable future, they can make other decisions such as whether or not layoffs are called for, if the design of the product needs to be reconsidered, if the business needs to be taken in another direction, and other decisions regarding corporate strategy. The ability to forecast future events with some degree of accuracy is necessary not only for the operation of the organization itself but also for all the members of the supply chain. The same knowledge about the demand for widgets that will affect the widget manufacturer will also affect the organizations providing raw materials or component parts, storing parts or products, delivering products, and selling them to the customer. For these and other reasons, it is important for successful business operations that forecasts be made and that these forecasts be as accurate as possible. With good forecasts, an organization is able to make decisions, develop strategy, and plan for the future.

Deterministic Variables that Affect Operations & Profitability

There are a number of deterministic variables for which there are specific causes or determiners that can affect the operations and profitability of a business. A trend is the persistent, underlying direction in which something is moving in either the short, intermediate, or long term. Identification of a trend allows managers to better plan to meet future needs. For example, a market trend for an increasing reliance on electronic gadgets may mean that a business needs to rethink its strategy of increasing its emphasis on manual tools. Business cycles are continually recurring variations in total economic activity. These expansions or contractions of economic activity tend to occur across most sectors of the economy at the same time. For example, several years of a boom economy with expansion of economic activity (e.g., more jobs, higher sales) are frequently followed by slower growth or even contraction of economic activity. Seasonal fluctuations are changes in economic activity that occur in a fairly regular annual pattern. Seasonal fluctuations may be related to seasons of the year, the calendar, or holidays. In most situations, for example, it would be unwise for a retail store to hire holiday workers on a permanent basis rather than only for the holiday shopping period.

Determining the Technique for Forecasting

There are many statistical techniques that can be used in forecasting. However, each is not equally applicable in every situation. The first decision a manager needs to make in choosing a forecasting method is to determine whether or not there are sufficient data available for quantitative analysis. If there are not, qualitative methods must be used. On the other hand, as shown in Figure 1, if there are sufficient data available, there are a number of techniques from which to choose. In order to choose the best technique for the data available, several questions must be asked. First, it must be determined whether or not there is useful knowledge available concerning the relationships and associations between the various factors of interest for the forecast. If there are not, then the type of data available -- cross-section or time series -- is a determining factor in which analysis techniques are most appropriate. For cross-section data, it must be considered whether or not the forecast needs to assess policy options or otherwise choose between alternative courses of action. If not, quantitative analogies are the most appropriate tool. In this approach, managers or other experts identify analogous situations and these inputs are used to derive the forecast (e.g., to determine how many seats are needed in a movie theatre in a new development, one could look at average data from movie theatres in similar developments). If, on the other hand, the forecast will be used to make decisions between alternatives, a better approach would be to employ an expert system. These are decision support systems that utilize artificial intelligence technology to evaluate a situation and suggest an appropriate course of action. Expert systems develop rules for forecasting following the reasoning processes used by decision making experts.

Forecasting Techniques Using Time Series Data

If time series rather than cross-section data are available, other techniques are more appropriate. If there is a good knowledge about the subject domain of the forecast, rule-based forecasting should be used. These approaches use an expert system that utilizes both expert domain knowledge and statistical techniques. If there is little or no knowledge about the domain, however, other options are available. Extrapolation techniques analyze times series data in an attempt to forecast future events (see below). Neural nets are an approach to artificial intelligence in which computer processors are connected in a way similar to the connections between neurons. These systems are able to learn through trial and error. Data mining can also be used for this type of situation. In data mining, large collections of data are analyzed to establish patterns and determine previously unknown relationships.

If there is a good knowledge of the relationship between variables and the future which is being forecast is unlikely to differ significantly from the past, extrapolation, neural nets, and data mining techniques are available. If, however, it is expected that the future events being predicted will differ significantly from the past, causal models or segmentation are better options for analyzing the data and making forecasts. In causal models, a combination of theory, research, and expert understanding of the domain are used to specify the relationships between variables and make a forecast. Regression analysis is one technique frequently used in this situation. However, although econometrics has been found to improve the accuracy of forecasts in this situation, the use of system dynamics has not.

Subjective & Qualitative Methods for Forecasting

In addition to quantitative methods for forecasting, there are also subjective or qualitative forecasting methods that are used by many managers. These are particularly useful when there are insufficient quantitative data for analysis or if a decision needs to be made quickly. These approaches are used when sufficient quantifiable data are not available for statistical analysis and are based on the manager's experience and intuition about a situation rather than on the application of mathematics or an attempt to reduce the situation to quantifiable terms. There are two major disadvantages of the use of qualitative techniques for forecasting. First, although some managers have good instincts and can make reasonable forecasts using subjective methods, quantitative methods are less dependent on the insights and experiences of one individual and use empirical, verifiable data. Second, because they are subjective, the results of subjective approaches are typically not reproducible because the variables cannot be quantified and applied to future situations.

This is not to say that qualitative techniques are not without merit. Experienced and insightful managers can take advantage of years of experience to extrapolate trends in ways that are still not possible through the use of quantitative techniques alone. In some instances, there are insufficient data to use quantitative techniques, necessitating the use of qualitative forecasting methods. Even when sufficient data are available, it is the human being who must decide which variables to include in the analysis and who must interpret the results of the forecast. Judgment is key to determining which data are relevant to the model. However, a statistical model or analysis cannot take every possible variable into account. Even if this was possible, spurious positive results would be seen due to the effects of probability alone. Expert judgments are essential to determine which inputs need to be considered during the forecasting process. In addition, in many instances, the determination of which analysis methodology is most appropriate to use for forecasting is a matter of judgment. There are a number of statistical techniques available for model building, analysis, and forecasting, many of which are closely related. The validity of the result depends heavily on correctly choosing the most appropriate analytical method. In addition, expert judgments can be helpful in understanding the situation and giving the manager insight regarding the parameters within which the data and subsequent analysis should be interpreted (e.g., "given the current economy and our market plan, we expect widget sales to rise by two percent over the next quarter"). As a result, quantitative and qualitative analyses are inseparable for most forecasts.

Applications

One frequently used statistical approach to forecasting is the analysis of time series data. These are data gathered on a specific characteristic over a period of time at intervals of regular length. Unlike ad hoc approaches to forecasting where it is impossible to tell whether or not the formula chosen is the most appropriate for the situation; time series analysis allows one to study the structure of the correlation of variables over time to determine the appropriateness of the model. The resultant model can be adjusted as needed to make it more representative of the real world situation.

Goal of Time Series Analysis

The goal of time series analysis is to build a model that will allow managers to forecast future needs. To do this, one must first specify the parameters of the model, including the degree of homogeneity in the time series and the order of the moving average and autoregressive components of the analysis. After the model has been specified, it is next estimated, frequently through nonlinear regression. The autocorrelation function is next examined using a simple chi-square test to determine whether the residuals are uncorrelated. The model must next be evaluated to determine its validity and whether or not it can be used to make accurate forecasts. Methods to do this include historical simulation starting at different points of time. Model building is an iterative process. The model can be refined as necessary to make it better fit the real world.

Techniques to Smooth out Fluctuation Effects

There are several types of techniques are used to smooth out irregular fluctuation effects in time series data. Naïve forecasting models are simple models that assume that the best predictors of future outcomes are the more recent data in the time series. Because of this assumption, naïve forecasting models do not consider the possibility of trends, business cycles, or seasonal fluctuations. Therefore, the naïve forecasting models work better on data that are reported more frequently (e.g., daily or weekly) or in situations without trends or seasonality. However, care must be taken since naïve model forecasts are often based on the observations of one time period so they can easily become a function of irregular fluctuations in data.

A second approach to smoothing time series data uses averaging models. These models help neutralize the problem of naïve models in which the forecast is overly sensitive to irregular fluctuations. In averaging models, data from several time periods are taken into account. In the simple average model, the forecast is the average of the values for a specified number of previous time periods. Moving averages, on the other hand, use both the average value from previous time periods to forecast future time periods, and update this average in each ensuing time period by including the new values not available in the previous average and dropping out the date from the earliest time periods.

Moving averages have the advantage of taking into account the most recent data available. However, it can be difficult to choose the optimal length of time over which to compute the moving average. Further, moving averages do not take into account the effects of trends, business cycles, and seasonal fluctuations. To help neutralize these problems, a weighted moving average can be used which gives more weight to some time periods in the series than to others.

A third approach to smoothing time series data is exponential smoothing techniques. These techniques use weight data from previous time periods with exponentially decreasing importance. Although all these approaches to time series modeling can be helpful for simple data sets, they do not account well for trends. However, there are several approaches available that can help managers forecast the influence of long-term changes in the business climate, including linear regression and regression quadratic models. However, the time series data cannot be influenced by seasonal fluctuations if these methods are to produce accurate forecasts. Otherwise, other techniques must be used. One of these is decomposition, in which the time series data are broken down into the four component factors of trend, business cycle, seasonal fluctuation, and irregular or random fluctuation.

Dealing with Spurious Data

One consideration that must be taken into account in the analysis of time series data is the possibility of spuriousness occurring as the result of error terms of the model being correlated with each other. This autocorrelation (or serial correlation) causes problems in the use of regression analysis because regression analysis assumes that error terms are not correlated because they are either independent or random. When this situation occurs, the estimates of the regression coefficients may be inefficient. Further, both the variance of the error terms and the true standard deviation may be significantly underestimated because of their effect. Autocorrelation also means that the confidence intervals and t and F tests are no longer strictly applicable. There are, a number of ways to determine whether or not autocorrelation is present in time series data including the Durbin-Watson test. Autocorrelated data can be corrected through techniques such as the addition of independent variables and by transforming variables.

Autoregression Analysis

Another approach to forecasting using time series data is autoregression. This is a multiple regression technique in which future values of the variable are predicted from past values of the variable. Autoregression takes advantage of the relationship of values in different time periods. In autoregression, one tries to forecast a future value of a variable from knowledge of that variable's value in previous time periods. The autoregressive approach can be useful for locating both seasonal and cyclic effects.

Integrated Techniques to Time Series Data Analysis

Another approach to analyzing time series data is to use mixed or integrated techniques that utilize both moving average and autoregressive techniques. For example, the autoregressive integrated moving average (ARIMA) model (also called the Box-Jenkins model) is an integrated tool for understanding and forecasting through the use of time series data. Although ARIMA models can be difficult to compute and interpret, they are powerful and frequently result in a better model than either the use of moving averages or autoregressive techniques alone. They can be used to determine the length of the weights (i.e., how much of the past should be used to predict the next observation) and the values of these weights.

Terms & Concepts

Artificial Intelligence (AI): The branch of computer science concerned with the development of software that allows computers to perform activities normally considered to require human intelligence. Artificial intelligence applications include the development of expert systems that allow computers to make complex, real world decisions; programming computers to understand natural human languages; development of neural networks that reproduce the physical connections occurring in animal brains; and development of computers that react to visual, auditory, and other sensory stimuli (i.e., robotics).

Autocorrelation: (also called serial correlation): A problem occurring over time in regression analysis when the error terms of the forecasting model are correlated.

Autoregressive Integrated Moving Average (ARIMA): An integrated tool for understanding and forecasting using time series data. The ARIMA model has both an autoregressive and a moving average component. The ARIMA model is also referred to as the Box-Jenkins model.

Data Mining: The process of analyzing large collections of data to establish patterns and determine previously unknown relationships. The results of data mining efforts are used to predict future behavior.

Deterministic Variables: Variables for which there are specific causes or determiners. These include trends, business cycles, and seasonal fluctuations.

Empirical: Theories or evidence that are derived from or based on observation or experiment.

Forecasting: In business, forecasting is the science of estimating or predicting future trends. Forecasts are used to support managers in making decisions about many aspects of the business including buying, selling, production, and hiring.

Independent Variable: The variable in an experiment or research study that is intentionally manipulated in order to determine its effect on the dependent variable (e.g., the independent variable of a type of cereal might affect the dependent variable of the consumer's reaction to it).

Model: A representation of a situation, system, or subsystem. Conceptual models are mental images that describe the situation or system. Mathematical or computer models are mathematical representations of the system or situation being studied.

Regression: A statistical technique used to develop a mathematical model for use in predicting one variable from the knowledge of another variable.

Strategy: In business, a strategy is a plan of action to help the organization reach its goals and objectives. A good business strategy is based on the rigorous analysis of empirical data, including market needs and trends, competitor capabilities and offerings, and the organization's resources and abilities.

Supply Chain: A network of organizations involved in production, delivery, and sale of a product. The supply chain may include suppliers, manufacturers, storage facilities, transporters, and retailers. Each organization in the network provides a value-added activity to the product or service. The supply chain includes the flow of tangible goods and materials, funds, and information between the organizations in the network.

Supply Chain Management: The process of efficiently connecting the parties in a value chain in order to reduce costs, improve customer service, develop the organization's knowledge base, increase efficiency, and create barriers to competitors. Supply management includes managing the flow of materials, information, and money within and between organizations in a supply chain.

Time Series Data: Data gathered on a specific characteristic over a period of time. Time series data are used in business forecasting. To be useful, time series data must be collected at intervals of regular length.

Bibliography

Armstrong, J. S. (2001). Forecasting. In Saul I. Gass, S. I. & Harris, C. M. (eds.), Encyclopedia of Operations Research and Management Science (pp. 304-310). New York: Wiley. Retrieved July 18, 2007, from EBSCO Online Database Business Source Complete. http://search.ebsco-host.com/login.aspx?direct=true&db=bth&AN=21891406&site=ehost-live

Arrfelt, M., Wiseman, R. M., & Tomas M. Hult, G. G. (2013). Looking backward instead of forward: aspiration-driven influences on the efficiency of the capital allocation process. Academy of Management Journal, 56(4), 10811103. Retrieved November 15, 2013, from EBSCO Online Database Business Source Complete. http://search.ebsco-host.com/login.aspx?direct=true&db=bth&AN=89878847&site=ehost-live

Armstrong, J. S. & Green, K. C. (2006, Sep). Select a forecasting method (selection tree). Retrieved July 19, 2007, from http://www.forecastingprinciples.com/selection%5Ftree.html

Black, K. (2006). Business statistics for contemporary decision making (4th ed. update). New York: John Wiley & Sons.

Krupnik, Y. (2013). Deploy business-specific predictive analytics in three easy steps. Supply & Demand Chain Executive, 14(3), 32-34. Retrieved November 15, 2013, from EBSCO Online Database Business Source Complete. http://search.ebscohost.com/login.aspx?direct=true&db=bth&AN=88842117&site=ehost-live

Ma, Y., Wang, N., Che, A., Huang, Y., & Xu, J. (2013). The bullwhip effect on product orders and inventory: a perspective of demand forecasting techniques. International Journal of Production Research, 51(1), 281-302. Retrieved November 15, 2013, from EBSCO Online Database Business Source Complete. http://search.ebscohost.com/login.aspx?direct=true&db=bth&AN=83404185&site=ehost-live

Nazem, S. M. (1988). Applied time series analysis for business and economic forecasting. New York: Marcel Dekker.

Suggested Reading

Armstrong, J. S. & Collopy, F. (1998). Integration of statistical methods and judgment for time series forecasting: Principles from empirical research. In Wright, G. & Goodwin, P. (Eds.). Forecasting with Judgment. New York: John Wiley & Sons.

Dauten, C. A. & Valentine, L. M. (1978). Business cycles and forecasting (5th ed.). Cincinnati: South-Western Publishing Co.

Di Giacinto, V. (2006). A generalized space-time ARMA model with an application to regional unemployment analysis in Italy. International Regional Science Review, 29(2), 159-198. Retrieved May 24, 2007, from EBSCO Online Database Business Source Complete. http://search.ebscohost.com/login.aspx?direct=true&db=bth&AN=20711879&site=ehost-live

Makridakis, S. & Wheelwright, S. C. (1982). Introduction to management forecasting: Status and needs. In Makridakis, S. & Wheelwright, S. C. (Eds.).The Handbook of Forecasting: A Manager's Guide. New York: John Wiley & Sons.

Morrell, J. (2001). How to forecast: A guide for business. Burlington, VT: Gower.

Nelson, C. R. (1973). Applied time series analysis for managerial forecasting. San Francisco: Holden-Day.

Wynne, B. E. & Hall, D. A. (1982). Forecasting requirements for operations planning and control. In Makridakis, S. & Wheelwright, S. C. (Eds.).The Handbook of Forecasting: A Manager's Guide. New York: John Wiley & Sons.