Robust estimators for the fixed effects panel data model

The presence of outlying observations in panel data can affect the classical estimates in a dramatic way. Nevertheless the common practice seems to disregard the problem. The aim of this work is to study robust regression techniques in the fixed effects linear panel data framework. Robustness of the procedures is investigated by means of breakdown point computations and simulation experiments. A distinction between outlying blocks and cells in a panel is made. To show the potential of robust panel data methods an empirical example on the response of the private sector behavior to fiscal policy is presented.status: publishe