[Mike Kraft]

Thanks a lot Joro Kolev. I appreciate your help. Yes, I actually missed that one.

After adding the missing bracket, the code runs. However, it gave me an error message

 Warning:  variance matrix is nonsymmetric or highly singular

and report all coefficients without z stats ?!

Calculating NLS estimates...
Iteration 0:  Residual SS =  15832.17
Iteration 1:  Residual SS =  15767.03
Iteration 2:  Residual SS =  15733.61
Iteration 3:  Residual SS =   15670.2
Iteration 4:  Residual SS =  15598.94
Iteration 5:  Residual SS =  15588.86
Iteration 6:  Residual SS =  15584.75
Iteration 7:  Residual SS =   15548.2
Iteration 8:  Residual SS =  15519.52
Iteration 9:  Residual SS =  15519.52
Iteration 10:  Residual SS =  15519.52
Iteration 11:  Residual SS =  15519.52
Iteration 12:  Residual SS =  15519.52
Iteration 13:  Residual SS =  15519.52
Iteration 14:  Residual SS =  15519.52
Iteration 15:  Residual SS =  15519.52
Iteration 16:  Residual SS =  15519.52
Calculating FGNLS estimates...
Iteration 0:  Scaled RSS =  190441.9
Iteration 1:  Scaled RSS =  190441.9
Warning:  variance matrix is nonsymmetric or highly singular

FGNLS regression
    
     Equation        Obs       Parms       RMSE      R-sq     Constant
    
1      f_varA      38118      5   .1090723    0.3325          k10
2      f_varB     38118      2   .0557011    0.0426          k20
3      f_varC      38118      2   .0059518    0.0147          k30
4      f_varD      38118      2   .3630655    0.4728          k40
5      f_return    38118     15   .5102272    0.0862         kr10
    

    
Coef.   Std. Err.      z    P>|z|     [95% Conf.    Interval]
    
/k10   -.0066366          .        .       .            .    .
/k11    .5234186          .        .       .            .    .
/k12    .0326297          .        .       .            .    .
/k13    .6014055          .        .       .            .    .
/k14   -.0241352          .        .       .            .    .
/k20   -.0151926          .        .       .            .    .
/k22    .1946063          .        .       .            .    .
/k30   -.0012101          .        .       .            .    .
/k33    .1117971          .        .       .            .    .
/k40    .2061318          .        .       .            .    .
/k44    .6831442          .        .       .            .    .
/theta1    1.149537          .        .       .            .    .
/kr10    5.787479          .        .       .            .    .
/kr11    .4492162          .        .       .            .    .
/kr12     1.02304          .        .       .            .    .
/kr13    4.703127          .        .       .            .    .
/kr14     .033854          .        .       .            .    .
/theta2    .3995938          .        .       .            .    .
/kr20    2.669952          .        .       .            .    .
/kr22   -1.796374          .        .       .            .    .
/theta3     1.65004          .        .       .            .    .
/kr30    -4.40338          .        .       .            .    .
/kr33   -2.815232          .        .       .            .    .
/theta4    .3020648          .        .       .            .    .
/kr40   -1.247302          .        .       .            .    .
/kr44    .2333843          .        .       .            .    .

I then thought to drop the zero value of one of the variables (that has lots of zeros), but then the output showed some z stats for some parameters but not all?!

I think the code struggles in estimating the intercepts in the f_return equation (i.e. kr10, kr20, kr30..)

Is there any way to solve this problem?

Thanks a lot