Hi everyone,


I am trying to make a RIF regression analysis using an inequality index that uses two Gini indexes from different variables and I want to be sure to obtain the same results with different procedures. So, I perform RIF regression in two ways. 1. Calculate a RIF variable using rifvar() function of egen and then run a regression with the RIF variable and the covariates. 2. Run rifhdreg with the dependent variable and the covariates, specifying specify the RIF statistic used as outcome . However, I do not understand why I obtain different estimates from these two procedures. Here is my example code:


Code:
webuse nlswork, clear
gen wage=exp(ln_wage)

** First way

egen rif_gini=rifvar(wage), gini
regress rif_gini age grade union tenure hours wks_work


** Second way

rifhdreg wage age grade union tenure hours wks_work, rif(gini)
I want to know if it is correct to make the analysis only with rifvar() and regress.

Thanks in advance.

Nicolás Mancera