Invertibility conditions for exponential smoothing models
journal contributionposted on 05.06.2017, 06:10 by Hyndman, Rob J., Akram, Muhammad, Archibald, Blyth
In this article we discuss invertibility conditions for some state space models, including the models that underly simple exponential smoothing, Holt's linear method, Holt-Winters' additive method and damped trend versions of Holt's and Holt-Winters' methods. The parameter space for which the model is invertible is compared to the usual parameter regions. We find that the usual parameter restrictions (requiring all smoothing parameters to lie between 0 and 1) do not always lead to invertible models. Conversely, some invertible models have parameters which lie outside the usual region. We also find that all seasonal exponential smoothing methods are non-invertible when the usual equations are used. However, this does not affect the forecast mean. Alternative models are presented which solve the problem while retaining the basic exponential smoothing ideas.