A note on the small sample bias of the Gini concentration ratio