Dear All,

This is my very first post on Satatlist. If anything is inappropriate, please let me know.

I would like to estimate a special probit model with one of the coefficients to be negative. After reading the FAQ at

https://www.stata.com/support/faqs/s...l-constraints/

I tried to write my own codes as the followings

Code:
program mynormal_b
        version 16
        args todo b lnf
        tempname a
        tempvar xb mu
        mleval `a' = `b', eq(1) scalar
        mleval `xb' = `b', eq(2)
        quietly {
                generate double `mu' = `xb' + `a' * $x2
                mlsum `lnf' = $ML_y1 * ln(normal(`mu')) + (1-$ML_y1) * ln(1 - normal(`mu'))
        }
end

global x2 _d_Mean_unitpricedluckiness_1

ml model d0 mynormal_b (a: ) (xb: luckiness_d2 = b_gender_d2 b_birthyear b_local br_child br_old br_parent br_move br_invest br_commute)

ml search, repeat()

ml maximize
When I run this, I got the error message "could not find feasible values r(491);"

I have two questions regarding this (I know I have not placed any constraints)

1. Even without any constraints, the model cannot be estimated. What Should I do?

2. What is the best way to re-parametrize/code the constraint (so the coefficient for variable _d_Mean_unitpricedluckiness_1 is negative)

Any help is much appreciated.

Regards,

Leo