Forum: open-discussion
Monitor Forum | 
Start New Thread
| RE: Error with the covariance matrix of the maximum likelihood estimates [ Reply ] By: Arne Henningsen on 2019-06-21 04:42 | [forum:46812] |
| It's seems that your R code does *not* fail due to an *error*! The model is estimated and you obtained estimates of all parameters. However, the sfa() *warns* you about a non-positive semidefinite covariance matrix of the maximum likelihood estimates. This could have many different reasons... | |
| Error with the covariance matrix of the maximum likelihood estimates [ Reply ] By: Yakira Fernández on 2019-06-20 17:40 | [forum:46806] |
|
Dear Prof. Arne Henningsen: I have an unbalanced panel. I`m trying to estimate a translog funtion with "sfa" in R, but I can`t because I get this error : Warning message: In sfa(logvts ~ l1logemp + l1logat + l1logpfa + I(0.5 * l1logemp^2) + : the covariance matrix of the maximum likelihood estimates is not positive semidefinite Please, could you help me?. I don`t know what to do. Excuse my English. Best regards, Yakira |
|
