Johansen cointegration test

[nooob]

Dear all,

I'm unfamiliar in the field of econometrics, but need the cointegration test for my thesis. So i was hoping on your help in telling me whether i'm on the right track and how to interpret my results. So i have 6 time series (998 daily data) and i need to define their effect on each other. I have tested them all using ADF and they were all non stationary in level and stationary in first difference.

My next step was to define the number of lags for Johansen CT. I did this by Estimate Var (Unrestricted) and then Lag Lenght Criteria with Lag specification 30 because of daily data. There i got some mixed results: 24 lags at LR, 3 lags at FPE, 3 lags at AIC, 1 lag SC and 1 at HQ. I chose to use AIC instead of LR value, because it is closer to SC and HQ, while LR is very far off (is this correct?).

Down here i have the results, but how do i interpret them, i mean what are these results telling me? And what does it mean to have different conclusions in Trace and MaxEigenvalue? Is my next step the Granger Causality test and do i have use the same number of lags?

Thank you very much

[Pandelis]

You should generally chose methodologies you are acquainted with or you are able to study and understand on your own.
Anyway:

FIRST you check whether your series have a unit root. If less than two of your variables are non-stationary (they have a unit root) then there cannot be co-integration.

Then (given you have at least 2 variables with a unit root) you should start with the maximum lag length and finally choose the lag length that minimizes the value of a model selection criterion (e.g. Akaike).

THEN you check if the model performs well in terms of homoskedasticity, no-serial correlation and Normality.

THEN you test for Granger causality (weak exogeneity).

THEN you perform co integration tests, which the results you present.

IF the results are the same, that means that you have co-integration, since the tests indicate the existence of a co-integrating vector which actually express the long-run relationship between your variables.