Testing Trend Stationarity Against Difference Stationarity in Time Series

  • Martin Moryson
Part of the Contributions to Statistics book series (CONTRIB.STAT.)

Abstract

The large sample tests exposed so far are not only useful for testing the constancy of regression coefficients over time, they can also be employed to test the stationarity of time series against the alternative of being integrated of order one. Most unit root tests are based on the Dickey-Fuller model:
$$ H_0 :y_t = \gamma _0 + \gamma _1 t + y_{t - 1} + u_t $$
(6.1)
vs
$$H_1 :y_t \, = \,\gamma _0 \, + \,\gamma _1 \,t\, + \rho y_{t - 1} + u_t,\,\left| \rho \right|\, < \,1, $$
where {ut} is some stationary process.

Keywords

Covariance Autocorrelation 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Physica-Verlag Heidelberg 1998

Authors and Affiliations

  • Martin Moryson
    • 1
  1. 1.WiesbadenGermany

Personalised recommendations