I have 2 such variables in my var model that exhibit the above given condition. Should i do something special in my Var model.
Different methods give different break dates. If we should be concerned about these breaks then which method is reliable?
Statistical inference is uncertain whether it involves the adf or the pp or the zivot andrews test for stationarity test. . If I were you I would adopt the zivot andrews of a structural break and continue with the analysis. This does not mean that this method is superior to the other methods. No method is always better than any other method. The prevalent condition in any situation may dictate which method to adopt.
Because of presence of breaks in the variables, stationarity ZA test is the best fit. As far as analysis part is concerned the best would be Markov -Switching VAR which will endogenously takes care of the break
- Badges
- Science topic