A Survey of New Dynamic Macro Econometric Models:An Automatic General-to-Specific Approach (in Persian)

Econometric methodologies can be evaluated based on several criteria, including consistency with economic theory, empirical robustness, forecasting performance, and policy relevance. Traditional theory-driven models often fail to account for non-stationarities in economic data—such as location shifts induced by structural changes—resulting in poor forecasting performance, unreliable policy simulations, and, at times, mathematical inconsistencies. This paper reviews the limitations associated with these evaluation criteria and introduces the Autometrics approach as a comprehensive model selection framework. Autometrics integrates theory-driven and data-driven strategies, often resulting in more candidate variables (N) than observations (T). The approach allows for the inclusion of theoretically relevant variables, dynamic adjustments, nonlinearities, deterministic and stochastic trends, and indicator saturation (IIS, SIS) to account for structural breaks and exogeneity, while preserving the integrity of theory-model parameter estimates. By employing a multi-path model selection algorithm, Autometrics identifies a congruent, parsimonious, and encompassing model through rigorous diagnostic testing of the reduction process.