An Integrated Model of the Data Measurement and Data Generation Processes with an Application to Consumers' Expenditure

The author shows that substantial reductions in the mean square error of preliminary vintages of data on consumers' expenditure can be obtained from the state space approach, providing further evidence that preliminaryvintages are not efficient forecasts of the final vintage.

Abstract

An integrated model is defined as one which not only models the data generation process (DGP) but also models the data measurement process (DMP). A natural framework for such an integrated model is the state space approach, with the optimal combination of preliminary vintages of data and predictions from the DGP model being obtained by application of the Kalman filter. We show that substantial reductions in the mean square error of preliminary vintages of data on consumers' expenditure can be obtained from this approach. This provides further evidence that preliminary vintages are not efficient forecasts of the final vintage.