Decomposition of non linear: Gomulka and Stern (1990)

[Afroza Alam]

Hello Everyone,
I am writing my masters thesis.
I am working with labour force survay data. I want to do Gomulka and Stern (1990) decomposition analysis (it's a pobit model). It has been used in the following paper.

Gomulka, J., and Stern, S., (1990), “The Employment of Married Women in the United Kingdom 1970-83”, Economica, 57: 171-199.


I am using STATA Verson 14. How i should do Gomulka and Stern decomposition (1990) in STATA. It will be really great if someone can help about this.

Thank you in advance

[Jean-Claude Arbaut]

Apparently a kind of Blinder-Oxaca decomposition. There are two articles in Stata Journal vol 8 no 4 (2014) :

* Ben Jann, The Blinder–Oaxaca decomposition for linear regression models, and the accompanying oaxaca package (type -ssc describe oaxaca- in Stata)
* Mathias Sinning, Markus Hahn, The Blinder–Oaxaca decomposition for nonlinear regression models, and the accompanying nldecompose package (type -net sj 8-4 st0152- in Stata)

You want the nonlinear one.

Hope this helps,

Jean-Claude Arbaut

[Afroza Alam]

Thank you so much..I will follow this

[Afroza Alam]

I am trying with nldecompose, but I have a confusion, I am using two separate data set like 2005 and 2010. and the same variables for 2005 and 2010. For example, education for 2005 and 2010, age for 2005 and 2010 and so on.
Now if I wand to see the changes happened between 2005 and 2010, If use nldecompose, do I need to combine them, or how should i do it.
nldecompose, just provide overall changes due to coefficient or characteristics. However, I also want to see their changes according to each independent variables. For example, how much education is responsible, how much age is responsible and so on.

Thank you ..