Journal of Official Statistics, Vol.18, No.4, 2002. pp. 511–530
Measures to Evaluate the Discrepancy Between Direct and Indirect Model-Based Seasonal Adjustment
Edoardo Otranto and Umberto Triacca
Abstract:In this article we deal with the problem of the evaluation of the discrepancy between direct and indirect seasonal adjustment. In a model-based framework, the direct seasonally adjusted series seems to be preferable, but a large discrepancy over the indirect seasonally adjusted series can cause confusion among the users. This is a crucial problem in respect of dissemination policy for the National Statistical Institutes. We propose a new approach to evaluate the size of the discrepancy, based on the idea that the two data generating processes of the alternative series (the direct and the indirect seasonally adjusted series) can be compared in terms of dissimilarity measures between RegARIMA models. A small dissimilarity implies that the difference between direct and indirect series is negligible. The procedure is performed in terms of classical hypothesis tests.
Keywords:Time series; ARMA; distance; forecastability.
Copyright © Statistics Sweden 1996-2018. Open AccessISSN 0282-423XCreated and Maintained by OKS Group