[Limdep Nlogit List] "maximum iteration" and "EFF command" in frontier model

Bo Kim bo_kim at live.com
Wed Sep 30 06:54:58 EST 2009


 

 

I run Battese and Coelli(1995) model in the stochastic frontier model.

 

I have two questions.

 

(1)

In the BC model, the command, ";Maxit = 100" does not work.

The maximum iteration always reach 51.

 

 

(2)

the command ";Eff = u" does not work, and thus I cannot create the variable 'u' after regression.

 

Are these some technical errors in Limdep, or Are these commands not allowed in the stochatic frontier model?

 

I coded it as follows.

 

sample ;all$
frontier ;Lhs=Y; Rhs = t, X1, X2;
  ;model = BC
  ;pds = T
  ;rh2 = one, X3, X4

  ;maxit = 100
  ;keep=y

  ;eff=u$


 

(3)

Since, BC model is not truely panel data model, I tried the following alternative using Truncated normal.

After regressing, u gives all the same value, so that I can't calculate Technical efficiency as " TE (it) = exp(-u(it)). 

Here, "(it)" is for panel specification.

 

 

sample ;all$
frontier ;Lhs=Y; Rhs = t, X1, X2;
  ;model = T
  ;rh2 = one, X3, X4

  ;keep=y

  ;eff=u$

 

I appreciate any replies.

Thank you.

 

-Bo

 

 

 

 
 		 	   		  
_________________________________________________________________
Insert movie times and more without leaving Hotmail®.
http://windowslive.com/Tutorial/Hotmail/QuickAdd?ocid=TXT_TAGLM_WL_HM_Tutorial_QuickAdd_062009


More information about the Limdep mailing list