Empirical modeling in dynamic econometrics (Q1083014)

The roles of evaluation, discovery, and theory in modeling economic time series are analyzed. Being reductions of data processes, models reflect economic behavior, filtered by the measurement system; the reductions entail a taxonomy of information sets and hence of selection criteria characterizing congruency. No method of model construction is necessarily valid, but search affects only research efficiency, not the intrinsic usefulness of a model. Moreover, critical evaluation remains justifiable, with predictive failure revealing the information content of tiny samples, and encompassing sustaining progressivity. Theory-models of starts and completions of private dwellings based on dynamized static equilibria and intertemporal optimization are contrasted. Empirical modeling illustrates the analysis.