Error when estimating SUREG with constraints (Stata 18)

[Abhishek Tripathy]

Hello,

This is my first post on Statalist. I apologize if my question does not fit the etiquette requirements relevant to this forum.

I am trying to run the following SUR sysem using the sureg command on Stata 18. The dataset consists of 8,645 observations, none of which have missing values. The equation involves 7 coefficient constraints, which are essential identifying restrictions.

global listm lnNm z1 z2 z3 z4  Nm Nf Nc lny lny_z1 lny_z2 lny_z3 lny_z4 lny 
global listf lnNf z1 z2 z3 z4  Nm Nf Nc lny lny_z1 lny_z2 lny_z3 lny_z4 lny 
global listc lnNc z1 z2 z3 z4  Nm Nf Nc lny lny_z1 lny_z2 lny_z3 lny_z4 lny 

*** Constraining the coefficients
constraint 1 [W_m]lny_Nf + [W_f]lny_Nf + [W_c]lny_Nf = 0
constraint 2 [W_m]lny_Nm + [W_f]lny_Nm + [W_c]lny_Nm = 0
constraint 3 [W_m]lny_Nc + [W_f]lny_Nc + [W_c]lny_Nc = 0
constraint 4 [W_m]lny_z1 + [W_f]lny_z1 + [W_c]lny_z1 = 0
constraint 5 [W_m]lny_z2 + [W_f]lny_z2 + [W_c]lny_z2 = 0
constraint 6 [W_m]lny_z3 + [W_f]lny_z3 + [W_c]lny_z3 = 0
constraint 7 [W_m]lny_z4 + [W_f]lny_z4 + [W_c]lny_z4 = 0

sureg (W_m $listm) (W_f $listf) (W_c $listc), constraints(1 2 3 4 5 6 7)

After running the code, the error message suggests that there are insufficient observations:

I am in a bind now as I have cross checked with the published code that estimates a similar SUR system, which otherwise runs fine. I am not able to place what could be going wrong with what I am doing.

Request you to help me surmount this problem.

Thank you

[George Ford]

does it run without the constraints?

[Abhishek Tripathy]

Thanks for responding. I tried running the system without the constraints. Still does not run. Returns the same error message as before. Please suggest where this could be going wrong?

[George Ford]

Looks like a collinearity issue. A data issue.

get a correlation matrix on all your Xs.

or, run the model with one variable and add one at a time until it fails.

[Abhishek Tripathy]

Thanks for responding. It was indeed a data issue. It seems I had a lot of missing values in one of my dependent variables

W_c

, due to faulty merging. Now that I have cleaned it up, the equation runs as expected even with constraints. Thank you for drawing my attention to data issues.