Casual Forecasting Methods:

Quite frequently, especially for short-and medium term forecasting, this is convenient to assume that the demand producing process will remain stable. A forecasting model can then be developed that associates demand to some variable believed to cause changes in the observed demand level. For this, the following three steps are followed.

Identify One or More Variables that can be Assumed to Influence Demand

The type and number of variables to be considered are decided on the basis of desired degree of accuracy in the forecast, and the experience and insight of the forecaster into the case. Examples of variables considered for many items are GNP, disposable income, rate of births, percent literacy, employment, etc.

Select the Type of Relationship among the Demand and the Variables that Influence Demand

For ease in computations, linear or logarithmic relations are generally selected.

Validate the Forecasting Model so that it Satisfies Both Commonsense and Statistical Tests

Most commonly used causal models are

1. Regression and correlation analysis models

2. Econometric models

3. Input-output models

4. System dynamics models.