Simultaneous Quantile Regression (sqreg) with percentiles not defined by dependent variable

[Maxence Morlet]

Hi all,

Here are my data (cross-sectional administrative data):

input byte(OverallMarkES5010950061Yi OverallMarkES5010850060Xi SCTS)
57 56 1
70 75 0
46 50 1
55 59 0
62 74 1
52 46 0
67 64 1
63 43 1
61 73 0
70 62 0
72 66 1
68 70 1
76 77 1
72 70 1
61 74 0
70 71 1

I want to run the following regression:

sqreg OverallMarkES5010950061Yi SCTS, quantile(0.25 0.50 0.75) reps(1000)

However, if I do that, the quantiles will be taken from the distribution of the dependent variable, here OverallMarkES5010950061Yi.

I would like to define quantiles based on the distribution of the variable OverallMarkES5010850060Xi and then run the simultaneous quantile regression above. OverallMarkES5010850060Xi is not included in the regression.

Is this possible? If so, could anyone please let me know how?

[Joao Santos Silva]

Dear Maxence Morlet,

Maybe I am missing something (it has been a long day!), but I do not think that can be defined as a quantile regression. A quantile regression estimates the conditional quantile of the dependent variable just like OLS estimates the conditional mean. So, I do not think that it makes sense to use a different variable to define the quantiles, but I'll be happy to be corrected.

Best wishes,

Joao