st: Clustered robust standard errors in simultaneous equations system

["Konstantinidi, Theodosia (konstat1)" <t.konstantinidi@lancaster.ac.uk>]

Dear all,

I am trying to estimate a simultaneous equations system with clustered robust standard errors and test cross-equation restrictions using robust Wald tests. 'Sureg' works fine but it does not allow the - cluster() - option. I need to use 'mysureg' which allows clustering, but I get the message: "convergence not achieved". Many standard errors are missing. When I do not use the 'iterate' option, it seems that the estimation procedure continues indefinitely. The same happens when I use the - difficult- option. Any suggestions will be greatly appreciated. 

mysureg (y1 x1 x2) (y2 x1 x2) (y3 x3 x4 x1 x2), iterate (40)

 

Fitting constant-only model:

Iteration 0:   log likelihood =  33565.451  

Iteration 1:   log likelihood =  33565.451  

 Fitting concentrated model:

 Iteration 0:   log likelihood =  33565.451  

Iteration 1:   log likelihood =  105661.45  

Iteration 2:   log likelihood =  105661.45  

 Fitting full model:

 Iteration 0:   log likelihood =  105661.45  (not concave)

Iteration 1:   log likelihood =  270586.26  ....

Iteration 38:  log likelihood =  2427782.5  (backed up)

Iteration 39:  log likelihood =  2427782.5  (backed up)

Iteration 40:  log likelihood =  2427782.5  (not concave)

convergence not achieved

 

Thank you,

Theodosia 

 


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/