Cointegrating Polynomial Regressions: Fully Modified OLS Estimation and Inference
Abstract: This paper develops a fully modified OLS estimator for cointegrating polynomial regressions, i.e. for regressions including deterministic variables, integrated processes and powers of integrated processes as explanatory variables and stationary errors. The errors are allowed to be serially...| Link(s) zu Dokument(en): | IHS Publikation |
|---|---|
| Hauptverfasser: | , |
| Format: | IHS Series NonPeerReviewed |
| Sprache: | Englisch |
| Veröffentlicht: |
Institut für Höhere Studien
2011
|
| Zusammenfassung: | Abstract: This paper develops a fully modified OLS estimator for cointegrating polynomial regressions, i.e. for regressions including deterministic variables, integrated processes and powers of integrated processes as explanatory variables and stationary errors. The errors are allowed to be serially correlated and the regressors are allowed to be endogenous. The paper thus extends the fully modified approach developed in Phillips and Hansen (1990). The FM-OLS estimator has a zero mean Gaussian mixture limiting distribution, which is the basis for standard asymptotic inference. In addition Wald and LM tests for specification as well as a KPSS-type test for cointegration are derived. The theoretical analysis is complemented by a simulation study which shows that the developed FM-OLS estimator and tests based upon it perform well in the sense that the performance advantages over OLS are by and large similar to the performance advantages of FM-OLS over OLS in cointegrating regressions.; |
|---|