Abstract:
|
In this talk, we present a unit root testing using a self-normalization approach. A new unit root test statistic is presented with a slight modification to the Phillips-Perron statistic and using the self-normalization. The advantage of using this new statistic is i) the limiting distribution is pivotal and thus can be tabulated, ii) the computation of the statistic is simple, and iii) it does not involve any user chosen parameters. The size accuracy is shown asymptotically and in finite samples through simulations.
|