We propose new unit root tests using stationary instrumental variables in the framework of the Dickey-Fuller (DF) regression. The most noteworthy feature of the suggested tests is that they are free of nuisance parameters. Under the null hypothesis, the proposed test statistic converges to the standard normal distribution regardless of various types of linear deterministic trends or structural breaks in the time series.
Mathematics Subject Classification: 91B84; 62M10
Keywords: Unit Root Tests; Dickey Fuller; Instrumental Variables; Standard Normal Distributions; t-test