A0788
Title: Fully modified OLS estimation of seemingly unrelated cointegrating polynomial regressions with common regressors
Authors: Fabian Knorre - TU Dortmund University and Statkraft (Germany) [presenting]
Martin Wagner - University of Klagenfurt, Bank of Slovenia and Institute for Advanced Studies, Vienna (Austria)
Abstract: Two fully modified OLS-type estimators are developed for systems of seemingly unrelated cointegrating polynomial regressions (SUCPRs), which contain common integrated regressors. Commonly integrated regressors refer to integrated regressors appearing in more than one regression equation of the system. It is shown that the developed fully modified OLS-type estimators have nuisance parameter-free limiting distributions, even in the case of regressor endogeneity and error serial correlation. A Wald-type test for testing the null hypothesis of linear restrictions on the parameter vector and additional specification tests based on augmented and auxiliary regressions are provided. A group-wise pooling framework is also provided to increase estimation efficiency by estimating a subset of coefficients via cross-sectionally pooled regressors. The performance of the estimators and tests in finite samples is evaluated by means of a simulation study. Finally, the proposed methods are illustrated with an estimation of the money demand of Euro area countries within a system of equations.