Joao: I couldn't find any clear, earlier references in your paper for the quantile case. As you point out, Bhattacharya (2005) doesn't really do it. I'd like to know what the first reference to clustering with quantile regression is because I didn't include any references in 2e of my MIT Press book. I'm (supposed to be) working on 3e and I should fill that in if I missed something.
I agree that it's good to prove consistency. My thought was that, with fixed T, there is nothing much beyond the cross-sectional case. The lack of smoothness can be handled for each t, and then clustering the scores evaluated at the truth is just the usual clustering argument. But I should look at the proof in your paper.
I agree that it's good to prove consistency. My thought was that, with fixed T, there is nothing much beyond the cross-sectional case. The lack of smoothness can be handled for each t, and then clustering the scores evaluated at the truth is just the usual clustering argument. But I should look at the proof in your paper.
Comment