As I have previously said, I am a PhD student in economics, at the Linnéuniversitetet.
My field of study is economics of labour, and the focus is on ethnic discrimiantion. I am also interested in child labour, and other kinds of discrimination (disability and gender).
At the moment I am almost done with the first paper. It is about the different distribution in the unobservable characteristics of the applicants to a vacancy (the data are from two correspondence tests). The model I am using is what I have called the Neumark decomposition (link below). My supervisors and I claim that the measure of the taste based discrimination (estimated with a heteroskedastic probit model) can be decomposed in two elements:
– taste based discrimination depurated from the effect of the different distributions in the unobservable skills
– second-moment statistical discrimination (i.e., the effect of the different distributions in the unobservable characteristics)
the abstract of my paper: “Neumark replication”
A number of different methodologies have been used to detect discrimination in the labour market. The Neumark decomposition is a new statistical technique which comes in hand in case of heteroskedasticity. We claim that it allows the deconstruction of the initial estimate of the discrimination into two parts: the taste based discrimination and the second-moment statistical discrimination. We applied it to data from two previous experiments (Carlsson & Rooth  and Carlsson ); we found that the actionable discrimination is actually underestimated, and that the second-moment statistical discrimination has a positive effect on the chances to receive a callback, when the requirements for a vacancy are low.
I can send the draft upon request.
the Neumark paper is here