Dear Statalist members, has anyone ever tried to use unconditional quantile regression in stata (ivqte command; Firpo's two-spep 2007 estimator)? It seems not to work when the treament variable is an interaction term of two other dummy variables that are included as control variables in the model. In this case, the first step of the estimation does not run as the logit model perfectly predicts the outcome. Anyone has come across this issue before? I have in the end chosen to omit from the control variables one of interaction terms, but I am not sure if this would be a good solution. Tips on how to deal with this would be good. Thanks.