Structural time series modeling: a Bayesian approach

摘要

Alternative progressive strategies for specification of linear dynamic models are presented. The main theme is that specification is basically concerned with endowing the pure incidental case—i.e., the case of different moment for each observations—with progressively more structure. Linearity and exogeneity are successively introduced in that spirit, first at a global level, then at a sequential level. This latter case is also shown to provide fuller possibilities for specification by means of the concepts of innovation and noncausality. This analysis along with the prior specification is conducted within the limits imposed by the possibility of concluding the analysis with computable posterior distributions on the parameters of interest.