Cointegration Rank Test and Long Run Specification: A Note on the Robustness of Structural Demand Systems

Private demand systems provide a practical application for analyzing identification issues in cointegration analysis. The paper conducts Montecarlo simulation experiments of cointegrated demand systems by assuming non-separability of government consumption. This framework enables further to test the robustness of models under alternative empirical specifications in which the homogeneity restriction is assumed to hold. The results highlight that separability of utility function with respect to government spending and the over-inclusion of lagged dependent variables introduce important bias in identifying the long run demand system, while the model specification with homogeneity restriction perform better when the theoretical hypothesis is contained in the data.Length: 34 pages