A note on the relationship between top income shares and the Gini coefficient