MLE estimation with constrained parameter

[Mateus Maciel]

Dear users,

I am trying to perform a MLE with a constrained parameter, and I would like to constrain beta2 = 1.3. However, when I run the code, it treats beta2 as an unconstrained parameter.

Here is my code:

set seed 11
clear
set obs 500
matrix define means = (0,0)
matrix define covar = (1, 0.5 \ 0.5, 1)
drawnorm x z, means(means) cov(covar)
gen y=1+1.2*x+1.5*z+2*rnormal()

mlexp (-0.5*ln({sigma2})-0.5*(y - {beta0}-  x*{beta1} - z*{beta2=1.3})^2/{sigma2})

[Andrea Discacciati]

set seed 11
clear
set obs 500
matrix define means = (0,0)
matrix define covar = (1, 0.5 \ 0.5, 1)
drawnorm x z, means(means) cov(covar)
gen y=1+1.2*x+1.5*z+2*rnormal()


constraint define 1 /beta2 = 1.3 
mlexp (-0.5*ln({sigma2})-0.5*(y - {beta0}-  x*{beta1} - z*{beta2})^2/{sigma2}), constraint(1)

Your code specifies that the initial value for beta2 to be passed to the optimisation engine is 1.3.

[Mateus Maciel]

Thanks, Andrea Discacciati. I used your code, but it returned: "constraints not allowed".

[Andrea Discacciati]

It works with Stata v 15.1.

Which Stata version do you have?
If Stata 14, try with

constraint define 1 _b[/beta2] = 1.3

[Mateus Maciel]

Yes, it is the Stata 14. Unfortunately, it is still not working.

Thanks for the help, anyway.

[Andrea Discacciati]

Ok, last try:

constraint define 1  _b[beta2:_cons] = 1.3