Dear Suleyman,

I hope your are having time series data as you haven't mentioned about panel. You can use Johansen cointegration test to check for any long run relationship and if it exists you can apply VECM as your sample size is good enough.

Further you may use Granger causality test to check for any unidirectional or bidirectional causality running from one to another variable.

Thank you,

Suraj Sharma, PhD

Thank you very much Dr Suraj, I really appreciate your answer,

Yes am dealing with time series data, and about estimating a model, should I apply DOLS or OLS

Use Johansen Cointegration and then VECM. Vecm is a model itself.

Thank you very much Mousumi Bhattacharya

Dear Suleyman,

VECM itself is a model as this is a regression model for time series analysis. This is called vector error correction model.

As your all variables are integrated of order 1, the best method is to estimate VECM after cointegration test. It will produce long and short run coefficients which will help is establishing appropriate relationship among variables of interest.

Thank you