MLE Code

[Karen Arulsamy]

Hi everyone,
I am replicating this code as follows:

clear all
webuse set "https://rlhick.people.wm.edu/econ407/data"
webuse tobias_koop
keep if time==4
su ln_wage educ pexp pexp2 broken_home

regress ln_wage educ pexp pexp2 broken_home

program define ols_mle

args lnf xb log_sigma2

qui replace `lnf' =ln(1/sqrt(2*_pi*exp(`log_sigma2'))) - (($ML_y1 - `xb')^2)/(2*exp(`log_sigma2'))

end

ml clear

ml model lf ols_mle (ln_wage=educ pexp pexp2 broken_home) (log_sigma2


Could someone help me understand what the program define component is doing?
I am trying to tie this back to the likelihood function and log likelihood equations but having a hard time trying to reconcile the code and theory. I have attached the likelihood function and log likelihood function for the two variable case.

Thank you for your help with this.

[Karen Arulsamy]

Bumping this up in case someone has an idea how this works.

[FernandoRios]

Program define and args are simply part of the syntax of how Stata needs for you to define the objective function to be maximized, the parameters that will be estimated
the only part that connects directly with male is the line that starts with quietly replace