# Forecasting Market Share

## Article excerpt

A widely used measure of a company's marketing performance is the market share of the company's products. Market share ratio measure that allows the sales of a product for one company to be compared to total of the industry. Some companies make strenuous efforts to increase market share at the expense of competitors, because gain in the market share is generally considered one of the important measures of business success. Often a forecaster is asked to forecast market share rather than, or in addition to, actual sales.

Market share forecasts have become important planning and evaluation tools. Consequently, many forecasting models have been developed or modified to forecast market share. This paper looks at several models used to forecast market share.

MODELS OF FORECASTING MARKET SHARE

There are basically four types of market share forecasting models:

1. Time series models

2. Multiple linear regression models

3. Logit models

4. Conjoint analysis models

TIME SERIES MODELS

Time series models use past data to project into the future. They are one of the most frequently used models of forecasting. Nearly all business forecasting texts discuss time series so the mechanics are not covered in this paper. To predict market share, the procedure is to use market share values for the past periods to develop seasonal patterns as well as trends in the data. The market share trends are projected into the future. Trends of market shares are adjusted based on the seasonality of the period being forecasted. Generally time series market share forecasts have been accurate up to one year into the future for products that have two or more years of sales history, and when relative prices and advertising expenditures have been nearly constant. Given that the relative prices and advertising are constant, it is likely that time series models will be the best models of forecasting market share.

For products being sold in markets with unstable competitive environments where prices and advertising levels are, changing, time series models may not be appropriate. In such cases the models discussed below may produce the most accurate forecast of market share.

MULTIPLE LINEAR REGRESSION MODELS

When applied to market share, multiple linear regression models are often called response models because they explore the relationship between market share and changes in the marketing mix variables. Marketing mix variables are often divided into four categories:

1. PRODUCT

a. packaging

b. product improvements

c. product innovations

2. PROMOTION

b. sales calls

c. salesperson compensation

3. DISTRIBUTION CHANNELS

a. channel promotion

b. channel development

c. training

4. PRICING

a. price changes

b. coupons

Response models estimate the extent to which market share responds to changes in these variables. Regression coefficients predict the effect of changes in price, advertising, or their marketing mix on a firm's market share. The model is particularly helpful in separating (partitioning) the effects of the marketing mix variables on market share. For example, suppose that a firm combines a series of price increases with increases in advertising expenditures. In this situation, the simple correlations between market share and price, or between market share and advertising could be quite small. Consequently, a model that attempted to predict market share based on price or advertising alone would yield poor forecasts. Multiple regression, however, can separate the two effects so that managers can understand how market share changes with price changes (holding advertising expenditures constant) and advertising changes (holding price constant). Response models that use multiple linear regression help managers to understand the effects of market mix variables on market share in addition to making a forecast of market share. …

Search by...
Show...

### Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.