Checking for stationarity : which test to believe ? (dfgls, pperron, kpss)

[Geolien Bautistelli]

Hello everybody,

I am actually testing some variables in order to determine if there is some stationarity issues or not. I am using Augmented Dickey-Fuller, Philips Perron and KPSS Test. The problem is that the different test contradicts themselves. Augmented Dickey-Fuller highlight non-stationnarity while Philips Perron and KPSS Test highlight stationarity. Can someone explain why or which test to believe the most ?

Find below my stata outcomes :

[Geolien Bautistelli]

Hi again, I forgot to ask :

I know that for the augmented dickey fuller test, I should specify a number of lags. In order to do that I read that we can use "varsoc" command but I don't understand how does it work and how to read it, can someone help me with that ?

Below my varsoc outcome for the same variable as above: