Prediction of Final Data with Use of Preliminary and/or Revised Data
In the case of US national accounts the data are revised for the first few years and every decade, which implies that the data is not really the final data. The Monte Carlo integration non-linear filter proposed by Tanizaki (1991, 1993) and Tanizaki and Mariano (1994) was applied. The predictors were introduced as an application of the non-linear filtering problem. The US consumption function was taken as an empirical example, where the measurement equation was simply assumed to be linear or log-linear between the preliminary data and the final data while the transition equation was derived from the utility-maximization problem.
Journal of Forecasting
Mariano, Roberto S. and Tanizaki, Hisashi.
Prediction of Final Data with Use of Preliminary and/or Revised Data. (1995). Journal of Forecasting. 14, (4), 351-380. Research Collection School Of Economics.
Available at: http://ink.library.smu.edu.sg/soe_research/260