''Last-place Aversion'': Evidence and Redistributive Implications