Forecasting Paper

Abstract

Forecasts are extensively used to support business decisions and direct the work of operations managers. The two major types of forecasts are qualitative and quantitative. Within each of these types are multiple methods and models. Qualitative forecasts are based upon subjective data. Quantitative forecasts are derived from objective data. Both methods are not suitable for all situations and circumstances. Each has inherent strengths and weaknesses. The forecaster must understand the strengths and shortcomings of each method and choose appropriately. One example of forecasting is the United States Marine Corps use of forecasting techniques, both qualitative and quantitative, to predict ammunition requirements.

Forecasting Defined

Forecasting is "A statement about the future" (Anonymous, 2005). Operations management is designed to support forecasted performances and events. Specifically, operations managers allocate personnel, time, and resources in order to meet the demands of forecasts. The most successful companies achieve their results by assuming just such a proactive vice reactive posture.

While forecasting is widely used, it does not fit into a standard one size fits all model. Multiple proven methods and models exist. In this paper we will examine, compare, and contrast the two most commonly used methods, qualitative and quantitative forecasting. Lastly, as a case study, we will examine how the United States Marine Corps forecasts its fiscal year ammunition requirements.

Qualitative Forecasting

Qualitative forecasts are the least scientific. They are based exclusively upon subjective data, such as opinions and estimates (Aquilano, Chase & Jacobs, 2005). Within the realm of qualitative forecasting are multiple techniques and measures. These are: grass roots, market research, panel consensus, historical analogy, and the Delphi method (Aquilano, Chase & Jacobs, 2005).

Grass Roots

Grass roots can best be described as a bottom-up process. This method is predicated on the assumption that employees who closely interact with customers are best aware of the customers' desires (Aquilano, Chase & Jacobs, 2005). Inputs from the lowest level are progressively staffed to the highest level where the decision is ultimately made (Aquilano, Chase & Jacobs, 2005).

Market Research

Market research is performed by specialized companies who collect data on customer likes and dislikes regarding existing or proposed products (Aquilano, Chase & Jacobs, 2005). This data is then used to create forecasts (Aquilano, Chase & Jacobs, 2005).

Panel Consensus

Panel consensus forecasting employs a panel of individuals with varying degrees of experience, training, and seniority in order to produce a diverse, broad estimate of the future (Aquilano, Chase & Jacobs, 2005). The purpose of the diverse panel is to eliminate group think (Aquilano, Chase & Jacobs, 2005). Too often, however, group think results as the junior members feel pressured to endorse the senior members' views.

Historical Analogy

The historical analogy technique uses the past performance of similar products to forecast sales for new products (Aquilano, Chase & Jacobs, 2005). An example would be Mercedes Benz using sales data related to the BMW X5 sports utility vehicle (SUV) to forecast sales of their own ML-350 SUV.

Delphi Method

The Delphi method is a variation of the panel consensus. In order to avoid the intimidation of lower-level members, Delphi activity masks the identities of participants and grants equal weight to the inputs of all members (Aquilano, Chase & Jacobs, 2005). The Delphi method is a series of inputs. After each series, the panel reviews the inputs and each member updates his/her contributions. The process repeats itself until a consensus is reached (Aquilano, Chase & Jacobs, 2005).

Quantitative Methods

Qualitative forecasts are based upon calculations and are more accurate for long-term use. They are based exclusively upon objective data, such as past performance (Aquilano, Chase & Jacobs, 2005). Time series analysis is the most widely used quantitative forecasting method.

Time Series Analysis Forecasting

Time series analysis is a series of observations taken at regular intervals over a specified period of time (Anonymous, n. d.). The following are techniques of time series analysis: simple moving average, weighted moving average and simple exponential smoothing, exponential smoothing with trend, and linear regression (Aquilano, Chase & Jacobs, 2005).

Simple Moving Average

The simple moving average considers a series of data and uses past performance to predict future performance (Aquilano, Chase & Jacobs, 2005). It is an ongoing exercise. When new data becomes available, the oldest data is dropped from the series and forecasts are recalculated (Aquilano, Chase & Jacobs, 2005).

Weighted Moving Average

The simple moving average assigns equal weights to all periods considered (Aquilano, Chase & Jacobs, 2005). The weighted moving average allows the forecaster to assigns weights to each period considered (Aquilano, Chase & Jacobs, 2005). The only requirement is that the cumulative weights must equal 1. This method is particularly suitable for businesses with wide seasonal variance.

Simple Exponential Smoothing

Simple exponential smoothing accounts for the previous period's forecasting errors in order to more accurately develop the current forecast by applying a smoothing constant or response rate (Anonymous, n. d.). Exponential smoothing also eliminates the requirement for large amounts of historical data (Aquilano, Chase & Jacobs, 2005).

Exponential Smoothing With Trend

The addition of a trend correction accounts for errors in an exponentially smoothed forecast (Aquilano, Chase & Jacobs, 2005). Specifically, an additional smoothing constant, Ð', is used to account for the error between the actual performance and the forecast (Aquilano, Chase & Jacobs,