The issue you are facing is the scale of the coefficients which results from incorrect modeling. You have to divide the observations with a factor to scale down the observations so that the coefficients are not too large. Depending on the variable, you can also do log transformation (e.g lngdp). But before proceeding with the ARDL approach, ensure that there are no econometric issues that may arise from the data (multicollinearity, I(2) variables). Your observations must normally be N>30 for ARDL. I hope this helps. .
Thank you, Mpho Bosupeng sir. That was helpful. I was although able to resolve this by normalising the dataset as well.
Hi
You have to to again to be sure that the variables are correctly measured. Coefficients in thousands is too large.
Thank you.
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running