Stationary time series
Posted: Wed Apr 18, 2018 11:37 am
Hello,
Im new to Eviews and time series analysis. Im a student of Finance and I had to enroll a time series subject. We were told to find and analyse a nonstationary (which was clearly easy) and a stationary time series.
I found like 15 or 20 time series which seemed to be stationary (e.g. here). I use census x-13 to see whether there is seasonality. This particular time series, just like 10 other, is not stationary according to ADF test with NONE model. Its like there is pattern in me doing something wrong or being tremendously unfortunate, because i get these results with every single time series Ive tested.
ADF with C says its stationary, ADF with trend is usually not a good model since the trend is insignificant and ADF with NONE says its nonstationary.
I checked the correlogram of the SA'ed time series at level and there seems to be no order or any other generating process, so I assume that should confirm my theory.
Given the time I have spent trying to find a stationary series it seems like I will never be successful. I need to diagnose it (normality, LM test, ARCH test) and everything has to pass. After that I need to find applicable ARMA model and make simple predictions.
I dont want try SARIMA since we havent learned how so thats why I need to work with SA'ed time series, coz I heard ADF only work on SA'ed TS.
Would anyone here be able to help me, please? Im desperate. This is quite a task.
If either of you happend to have time series that is sure to be stationary, please do share with me.
Im new to Eviews and time series analysis. Im a student of Finance and I had to enroll a time series subject. We were told to find and analyse a nonstationary (which was clearly easy) and a stationary time series.
I found like 15 or 20 time series which seemed to be stationary (e.g. here). I use census x-13 to see whether there is seasonality. This particular time series, just like 10 other, is not stationary according to ADF test with NONE model. Its like there is pattern in me doing something wrong or being tremendously unfortunate, because i get these results with every single time series Ive tested.
ADF with C says its stationary, ADF with trend is usually not a good model since the trend is insignificant and ADF with NONE says its nonstationary.
I checked the correlogram of the SA'ed time series at level and there seems to be no order or any other generating process, so I assume that should confirm my theory.
Given the time I have spent trying to find a stationary series it seems like I will never be successful. I need to diagnose it (normality, LM test, ARCH test) and everything has to pass. After that I need to find applicable ARMA model and make simple predictions.
I dont want try SARIMA since we havent learned how so thats why I need to work with SA'ed time series, coz I heard ADF only work on SA'ed TS.
Would anyone here be able to help me, please? Im desperate. This is quite a task.
If either of you happend to have time series that is sure to be stationary, please do share with me.