collinearity in time series

[bouki_81]

I'm about estimating a regression equation with time series data, but the problem is that there is a correlation between two explanatory variables which is lead to collinearity, is there another way to deal with this problem rather than using instrument variable? cause the two variables are very important in the model that I'm want to estimate.

please, anyone has answer

best regards

[startz]

A least squares regression handles collinearity between the right hand side variables quite nicely.

[bouki_81]

A least squares regression handles collinearity between the right hand side variables quite nicely.

thanks, but the least squares regression is not useful in this case since the series are integrated at order one and the OLS is no longer valid to estimate series with I(1) process.

[startz]

Then the usual procedure would be first to test for cointegration. Then, if the series are cointegrated, estimate an error-correction model or some such.

[bouki_81]

thanks for your comment
but don't you think that running cointegration test then estimating the VEC model knowing that some explanatory variables has a collinearity problem could make the results bias and inefficient, and using time series approach with johansen cointegration test does not solve the problem of collinearity?

[startz]

Collinearity doesn't bias results or make them inefficient.

[bouki_81]

thanks startz
I'm really appreciate your advices.