Hi everybody!
Please, I´d thank you to help me with a doubt regarding forecasting based on a VECM.
I estimated a VECM model with 3 variables, there is just one cointegrating relation, and there is no problem with this estimation.
Now I want to forecast, and for that goal I have (from information of an international organization) the predicted values of 2 of the 3 variables included in my VECM. Therefore, what I want now is to predict that one remaining variable, using the estimation of the VECM, and the given external forecasts of the 2 other variables.
I think there must be the way to do it, but I don´t realize about it yet. I´m using STATA.
Any suggetions about this?
In advance, thank you all so much!!
See these links:
http://web.vu.lt/mif/a.buteikis/wp-content/uploads/2018/04/Lecture_07.pdf
https://www.researchgate.net/topic/VECM
https://www.analyticsvidhya.com/blog/2018/09/multivariate-time-series-guide-forecasting-modeling-python-codes/
It should be possible to re-formulate the VECM model as a state space model, with the two variables for which estimates are available as the measured variables and the third variable as the latent variable. This will allow you to use the linear Kalman filter for estimating the conditional mean and the conditional variance.
This link might be helpful to have a practical guide-
https://www.youtube.com/watch?v=Xu16T_Y8hUU
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression