Hi everybody! I have some doubts about the use of the Blinder-Oaxaca decomposition, and if somebody of you could kindly help me it would be great!
I am using the command 'oaxaca' in Stata to decompose the gender wage gap and understand the weight given by differences in observable characteristics (endowments) and the unexplained part (coefficients). I am using the default option given by Stata, i.e. threefold, and I have therefore also an interaction effect.
My doubt regards first the use of the reference group.. I am performing the following command:
xi: oaxaca lnhourlywage ( i.l5_1_mfr degreeontime i.degreegrade i.atheneumregion i.l1_1 i.voto_diploma_all i.eta_laurea_mfr i.tiplau_micro i.gruppo_micro i.domicile fulltime permanent i.l1_15 privatecourses privatesector weeklyhours i.l2_44 livingorigin doctoralstudies specializationschool masterafterdegree studyscholarship_workgrant stage trainership_practicum professionaltraining trainingcourse i.l5_4 mesi_i_lavoro durata_lavoro_mesi) if [people_over==0], by(sesso) threefold(reverse) relax
where 'sesso' is a dummy variable taking value 1 if the individual is female and 0 if it is male.
Looking at the 'help oaxaca' it states 'the threefold decomposition by default gives the viewpoint of group B'. If you want group A to be the reference, just use 'threefold(reverse)'. My output indicates sesso=0 as group 1 and sesso=1 as group 2. Am I doing it in the right way if I want to use male as the reference group and I therefore specify 'threefold(reverse)?
Moreover, I have some doubts on the interpretation of the results(attached files).. I thought endowments were the explained part considering differences in observable characteristics. My reasoning is the following one: if the endowment is negative and I have male as the reference group (by using threefold(reverse)), I think that females have worse observable characteristics.
However, I do not understand the reason why the sign does not change as I do not use threefold(reverse), i.e. if I use female as the reference group. shouldn't the endowments be positive in that case (so that the fact that men have better characteristics are again better and his is coherent since I have the same data)?
Another difficulty encountered is how to interpret the interaction result..
Thanks a lot in advance!!
Giulia
I am using the command 'oaxaca' in Stata to decompose the gender wage gap and understand the weight given by differences in observable characteristics (endowments) and the unexplained part (coefficients). I am using the default option given by Stata, i.e. threefold, and I have therefore also an interaction effect.
My doubt regards first the use of the reference group.. I am performing the following command:
xi: oaxaca lnhourlywage ( i.l5_1_mfr degreeontime i.degreegrade i.atheneumregion i.l1_1 i.voto_diploma_all i.eta_laurea_mfr i.tiplau_micro i.gruppo_micro i.domicile fulltime permanent i.l1_15 privatecourses privatesector weeklyhours i.l2_44 livingorigin doctoralstudies specializationschool masterafterdegree studyscholarship_workgrant stage trainership_practicum professionaltraining trainingcourse i.l5_4 mesi_i_lavoro durata_lavoro_mesi) if [people_over==0], by(sesso) threefold(reverse) relax
where 'sesso' is a dummy variable taking value 1 if the individual is female and 0 if it is male.
Looking at the 'help oaxaca' it states 'the threefold decomposition by default gives the viewpoint of group B'. If you want group A to be the reference, just use 'threefold(reverse)'. My output indicates sesso=0 as group 1 and sesso=1 as group 2. Am I doing it in the right way if I want to use male as the reference group and I therefore specify 'threefold(reverse)?
Moreover, I have some doubts on the interpretation of the results(attached files).. I thought endowments were the explained part considering differences in observable characteristics. My reasoning is the following one: if the endowment is negative and I have male as the reference group (by using threefold(reverse)), I think that females have worse observable characteristics.
However, I do not understand the reason why the sign does not change as I do not use threefold(reverse), i.e. if I use female as the reference group. shouldn't the endowments be positive in that case (so that the fact that men have better characteristics are again better and his is coherent since I have the same data)?
Another difficulty encountered is how to interpret the interaction result..
Thanks a lot in advance!!
Giulia
Attached Files
