An Empirical Comparison of Alternative Forecasting Methods as applied to the UK Foreign Holiday Market

Abstract

There are two distinct types of forecasting model in which past experience is used as an indicator of the future, and these may be termed “causal” models and “non‐causal” models. (An extensive discussion of these model types appears in Robinson and Wood and Fildes.) Non‐causal (naive) models simply extrapolate past history on the forecast variable and disregard those forces which caused the particular pattern for the time series. The object is to select the type of curve which provides the closest fit to a given historical series, and complex statistical procedures exist for carrying out this exercise. The great problem with forecasting by extrapolation is that it presupposes that the factors which were the main cause of growth in the past will continue to be the main cause in the future, which may be incorrect, and if this is the case the use of this technique will result in poor forecasts. If one considers sales of foreign holidays as an example, one realises that there may be significant changes in the variables affecting these sales, such as income changes, fare changes, and changes in exchange rates. In order to forecast sales of foreign holidays reasonably accurately it is therefore necessary to construct a causal model in which sales are explicitly related to the determining forces.

Citation

Witt, S.F. and Rice, R.A.C. (1981), "An Empirical Comparison of Alternative Forecasting Methods as applied to the UK Foreign Holiday Market", Managerial Finance, Vol. 7 No. 1, pp. 16-20. https://doi.org/10.1108/eb013480