A New Hausmann Type Test to Detect the Presence of Influential Outliers

In the presence of outliers in a dataset, a least squares estimation may not be the most adequate choice to get representative results. Indeed estimations could have been excessively infuenced even by a very limited number of atypical observations. In this article, we propose a new Hausman-type test to check for this. The test is based on the trade-off between robustness and effciency and allows to conclude if a least squares estimation is appropriate or if a robust method should be preferred. An economic example is provided to illustrate the usefulness of the test.