I have attempted to run OLS regressions using detrended data (as the data at levels are non stationary), but was wondering whether I should check for cointegration and if so, include the error correction term in my regressions? Well, that would be the case if I use differenced data, where the long run properties of my variables would be lost. But why one would choose to detrend over differencing is for ease of interpretation and also because it retains long run properties of the data.
I would appreciate any advise and suggestions, and if my question is unclear, please let me know.
Thanks in advance.
Adding to the above question...
Since I've used detrended data (and stationary, following that transformation), should I still test for cointegration and include an error term in my regressions?
- Badges
- Science topic
- Similar topics
- Correction