"convergence not acheived" qreg error

[abduelwahab hussein]

Hi,

I am attempting to run a quantile regression. The model has only dummies and interactions of dummy variables as predictor variables and a continuous variable as the outcome variable. I have tried running the model with and without clustered standard errors and am still getting a "convergence not achieved error".

For some reason, I am getting a "convergence not achieved" error. I believe it has something to do with the variance covariance matrix. What are some possible reasons I may be getting this error?

Thank you.

[Carlo Lazzaro]

abduelwahab:
the usual recipe is to run your regression from scratch, adding one predictor at a time and see when Stata starts gasping.

[FernandoRios]

A second recommendation.
the maximization algorithm of qreg is some times very sensitive to specification. If you have so many dummies and interactions, as you say, its possible that there is not enough information to identify the conditional quantiles for those dummies
I find that qrprocess (ssc) handle those problems better.