Forum: help
Monitor Forum | 
Start New Thread
| RE: Weighted SUR [ Reply ] By: Arne Henningsen on 2019-11-03 13:24 | [forum:47112] |
| I think so. I suggest that you test this with a single equation and compare the results obtained by systemfit() using the transformed variables with the results obtained by lm( ..., weights = ... ) using the original / non-transformed variables. If you get identical estimates and identical standard errors, the manual procedure will likely also work for systems of multiple equations. | |
| Weighted SUR [ Reply ] By: Stephen Clark on 2019-08-01 11:33 | [forum:46871] |
| I understand that the current implementation of SUR does not permit an observation weight. Here this is referred to http://r.789695.n4.nabble.com/Weighted-SUR-NSUR-td4670602.html . Would it be OK to manually weight (multiply) both the dependant and independent variables by the weight before the estimation to achieve the same outcome? | |
