Modelling a change of classification in economic time series data

The change of classification problem for economic sectoral time series data is examined by a approach. State space representations are proposed both for data reconstruction and modelling a change of classification. The Doran (1992) methodology of constraining the Kalman filter to satisfy time varying restrictions is applied to show how to handle both limited information and aggregation constraints. We explore the implications of this approach for what will be, perhaps, the most important change of classification in sectoral data: the new National Accounts for European Unification. Results of an experimental application to Italian Quarterly Accounts are provided