Cointegration analysis in the presence of structural breaks in the deterministic trend (Q2707872)

This research work is devoted to analyzing the cointegration in the Gaussian vector autoregressive models with broken linear trends and known break points, through the agency of reduced rank regressions, the combinations of least squares regression analysis and the canonical correlations analysis, and under appropriate rank hypothesis. The authors present important tests for the cointegration ranks of models having broken trends and broken levels and the simulations of the asymptotic distributions with the corresponding responses surfaces analysis. Various tests for linear restrictions on the slopes for the broken trends are given with asymptotically distributions. The paper also containts illustrations of the considered procedures using data for the inflation rates and exchange rate for Italy and Germany, being based on a very pertinent bibliography.