Estimation and Testing for Cointegration: A Spectral Regression Approach

Authors : Yılmaz Akdi

Pages : 95-111

View : 20 | Download : 17

Publication Date : 2013-07-15

Article Type : Research Paper

Abstract :A popular topic in the econometrics and time series area is the cointegrating relationship among the components of a vector autoregressive time series. The problem became important after the work of Engle and Granger (1987) and has been addressed by many authors: Johansen (1988), Stock and Watson among many others. Engle and Granger’s least squares method and Johansen’s conditional maximum likelihood method have received the most attention. These tests are routinely applied to economic time series because the notion of cointegration has a natural interpretation. Our method uses low frequency components of the cross periodogram to estimate the cointegration relationship between cointegrated time series. The method improves the results of ordinary least squares method proposed by Engle and Granger in some cases.
Keywords : Zaman Serileri, Eşbütünleşme, Periyodogram Ordinat, Spektral Regresyon