How Far Ahead Can an EWMA Model be Extrapolated?

The existence of uncertainty is a feature of the business world. Forecasting does not remove this uncertainty, but sets out to measure and minimize it. This paper suggests decomposing the forecast error into three components, namely the residual unexplained error, the error resulting from the estimation procedures of the model, and the errors due to the approximate nature of the model. The quantification of this division for a steady-state model, or exponentially weighted moving average (EWMA) process, is used as an illustration. The results obtained indicate the maximum safe limits for extrapolating this model into the future.