Can trend changes be forecasted?

Abstract

Essence of the method: sales forecasts can be made not on the basis of total turnover, but also the turnover components even if there are no statistics about them.
The method consists of the following steps:
1. Breakdown of the components of demand for a product from a particular point of view.
2. Establishment of time series on the purchases by the different target groups.
3. Establishment of trend values for the values in the different time series.
4. Extrapolation of trend values for the period of the forecast.
5. Determination of anticipated turnover by summing up the forecasted trend values.
In many cases representative surveys of the population must be conducted as it happened at the Hungarian Institute for Market Research in 1972 (study on washing machines). We divided demand into new purchases and replacements. We determined the time series for the two components by the yearly proportions of new purchases and replacements; these proportions were then projected for the entire retail trade turnover. Extrapolation of these sub-trends showed that the number of replacements will continue to increase rapidly, but the rate of decline in new purchases will slow down. This forecasting method turned out to be successful, after a short time the forecasted trend change has occurred.