Dear Md,

I wonder if you checked co-integration before running VECM regression. Such problem often occurs when the dependent variable and the independent variable do not co-integrate (in which case a VECM regression is not proper for the data).

Hope this helps.

Joy

Very right Joy.

Dear Joy Yang,

Thanks for your valuable comment. I check the co-integration and the rank was r=1.

Also when you take stationary variables as non-stationary, it will create some problems. Redo the Unit Root tests but mostly use DF-GLS test for better analysis. Don't just only rely on ADF and PP tests.

Introduce dummies and reestimate!!

the variable that i use are all stationary at 1rst difference and i have used VECM model for analysis but the model is not stable and the residuals are not normally distributed. how can i correct it ??