IDENTIFICATION OF UNOBSERVED COMPONENTS MODELS
Luiz Koodi Hotta · Journal of Time Series Analysis · 1989
Abstract. Unobserved components ARIMA models are common in time series applications. However, fitting models of this type leads to problems of model identification. In this paper we derive a methodology to check whether a proposed model is identifiable. We show that this kind of identification can be checked using the autocovariance generating function and/or the (pseudo‐)spectral generating function.