Hi, I am trying to estimate a model of charitable giving. I have a study where individuals are given x dollars to keep or donate. If they choose to donate some or all of it they are given a choice of charities. they can donate to only one charity. I would like to model the donation decision, charity choice and donation amount. I assumed multinomial charity choice can be followed by three different truncated regressions. it gives me a message of misspecification and fails to converge. What am I doing wrong? Here are the key constraints: One can keep it all or keep some or none. if choose to keep some or none you are donating. only one charity can be chosen. there are three charities. donations are between $0-X. so... amount kept + amount donated =X


cmp (donation_CharityA=X ) (donation_CharityB=X) (donation_CharityC=X) (charitychoice=X Z), ind($cmp_left $cmp_left $cmp_left $cmp_mprobit)

charitychoice is a catergorical variable (A B or C if someone chose to donate)

I also estimated the following model which converged.

cmp (donation_CharityA=X ) (donation_CharityB=X) (donation_CharityC=X) (anydonation=X Z), ind(anydonation anydonation anydonation $cmp_probit)