Francesco Bravo (University of York)
|Dates:||30 March 2017|
|Times:||16:15 - 17:45|
|What is it:||Seminar|
|Organiser:||School of Social Sciences|
Title: Semiparametric quantile regression with missing data.
Abstract: In this talk I discuss a general methodology to deal with missing data in semiparametric quantile regression models. The method is based on the inverse probability weighting (IPW) of the missing data mechanism and can be applied to a variety of empirically relevant situations. In the talk I consider two such situations: the response being subject to random censoring and some of the covariates being missing at random.
The estimation procedure that I propose combines the idea of IPW with that of backfitting iteration.The basic idea is to first obtain a consistent estimate of the infinite dimensional parameter and then use it to re-estimate the finite dimensional component of the model. As this second step requires undersmoothing an additional third step can be used to obtain a final estimate of the infinite dimensional parameter, should it be of interest. A computationally simple resampling technique can be used to consistently estimate the asymptotic covariance of the finite dimensional parameter estimator.
The asymptotic analysis of the response being subject to random censoring and of the missing covariates case requires different arguments, which will be detailed in the talk. Monte Carlo simulations and an empirical application illustrate the usefulness of the proposed general methodology.
Role: Reader in Economics
Organisation: University of York
Travel and Contact Information
Humanities Bridgeford Street