Oxaca Model Interpretation

[Patricia Ali]

Dear All,

Below is an except of my results from the Oxaca model

	Country A			Country B
Male	8.614***			9.714***
	(0.0566)			(0.0235)
Female	7.958***			9.085***
	(0.150)			(0.0539)
difference	0.656***			0.629***
	(0.160)			(0.0588)
Agregate Decomposition
	Endowment	Coefficient	Interaction	Endowment	Coefficient	Interaction
Total	0.165	0.342***	-0.00750	0.365***	0.256***	0.0202
Detailed Decomposition
Age	-0.0216*	-0.459	0.0650	0.213	2.134*	-0.283*
	(0.448)	(4.512)	(0.466)	(0.115)	(1.191)	(0.126)

I have about 10 explanatory variables of which some are significant. For simplicity I have only included Age. However, I am not sure how to explain the detailed decomposition and would appreciate guidance. My dependent variable is productivity.

Many thanks

[FernandoRios]

Hi Patricia,
First of all, my first suggestion is for you to recreate the OB decomposition by hand. That way you will know where every component comes from, and may facilitate your interpretation of the results.
Second. For your specific example, perhaps something like this could be said about the detail contribution of age.
Differences in age has contributed to reduce the gap of the total outcome differences. If women had were in average as old as men, the outcome gap would have been 0.0216 points larger in country A.
In country B, the differences in age do not seem to have a significant contribution on the observed gap, despite its magnitude.
Notice that without any other information, i do not know if age is possitively related to the outcome, who has a larger marginal effect respect to age (according to the OLS model) nor who is older (men or women). That information is needed to make better sense of your results.

For more specific recommendations, i would do a quick search on other research (in particular that related to your topic) who have done OB type of analysis, and follow them as guidelines to interpret your results.
HTH.

[Patricia Ali]

Thanks FernandoRios