Quantile regression with zero-inflated count data

[Daisy Pollenne]

Dear Statalist,

I am working with a zero-inflated count dependent variable. In my main models, I use the 'nbreg' command to account for the nature of the dependent variable.
As a second step, I wish to estimate quantile regression.
I understand that 'qcount' handles count data through jittering, but I am unsure how to best address the zero-inflation in the quantiles.

I have considered:

1. Using qcount directly, but worried about bias from zero-inflation
2. Two-part model approach, though unsure how to implement with quantile regression

Has anyone tackled this specific combination of zero-inflation and quantile regression for count data? Any suggestions for handling this empirically would be greatly appreciated.

Best regards,

Daisy Pollenne

[Joao Santos Silva]

Dear Daisy Pollenne,

I am a bit puzzled by what you describe. You say you have zero inflation and use nbreg to deal with that, but nbreg does not deal with zero inflation. Can you please explain why you say you have zero inflation and why you think that nbreg is a suitable way to deal with that? Also, can you please say a bit about what you are doing? With that information, I may be able to help.

Best wishes,

Joao