Will there be a problem of multicollinearity in a Vector Error Correction Model when we want to find the long and short run relationship between the prices of two brands of a same commodity, rice in my case?
Running a test of collinearity is a good choice, but the usual VIF does not give results for VECM in STATA. Do you know an alternative method to check for collinearity in the framework of Vector Error Correction Models?
It is not a serious problem. This relates to your data. If you know the variables co-linear, just test them jointly. You can interpret also jointly. To see each variable individually, you can drop it one by one. You can also combine the co-linear variable, based on the sign. Hope this helps
Thanks for your good advices. You are right without checking it is difficult to judge whether the problem of collinearity is acute or not. But, it is known that differencing reduce multicollinearity in time series models which is true in my case. However, I was not able to use Variance Inflation Factor (VIF) for testing collinearity. But, correlation matrix of coefficients of model showed strong correlation between two variables.
Do you know any other method to test for collinearity in time series models.