Good evening everybody.
I want to run Seemingly Unrelated Regressions (SUR) but the Variance-covariance matrix obtained is singular. But, in my estimations, I found that when I replaced the variance-covariance matrix (singular matrix) by its Jacobian matrix, I am able to run the estimations. I am sceptical about the validity of using this matrix as a kind of feasible generalized least squares.
So my questions are:
- Is it advised or not recommended to use the Jacobian matrix of residual variance-covariance matrix (that is singular) in the context of SUR estimations?
- Are the results, obtained from it, and the subsequent simulations (bootstrap) valid and relevant?
Thank you in advance.
Gratefully
not sure but it looks like this
there something that seems close in this
https://www.google.fr/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&ved=0ahUKEwic2bXG45LKAhXIChoKHTXrApsQFgglMAA&url=http%3A%2F%2Fds.data.jma.go.jp%2Ftcc%2Ftcc%2Flibrary%2Ftraining%2FStatistical_Method.ppt&usg=AFQjCNEpIKiBCN0Wg_42PFtXWXdghlbXBQ
hope it helps
Hi and thank you for your suggestion. It helps me a lot.
Also, in the document that you advised me, precisely in the slide 33, the author sent to the note for detail (In detail, please refer to the note). Can I have this note with you. Thank you in advance.
unfortunately I do not have this note
have a look there
http://ds.data.jma.go.jp/tcc/tcc/products/model/index.html
https://fr.wikipedia.org/wiki/Analyse_canonique_des_corr%C3%A9lations
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression