Nonlinearly testing for a unit root in the presence of a break in the mean

This paper deals with testing a time series with a structural break in its mean for a unit root when the break date is known. A nonlinear (with respect to coefficients) test equation is used, providing asymptotically efficient estimates. Finite-sample and quasi-asymptotic empirical distributions of the unit root test statistics are estimated, comparing them with those associated with the Perron-type equations. Asymptotic distributions of the nonlinear test statistics are found to be the Dickey-Fuller distributions. The nonlinear test proves to have more power than the test based on the linear model.