Changes

*Consider a two-dimensional case. Any policy <math>z_{i}</math> is characterized by coordinates <math>(x_i, y_i)</math>.

*In first stage we vote on <math>x_{i}</math>. and obtain policy equal to median voters bliss point x_m<math>x_{m}</math>. In second stage we vote on <math>y_i </math> and obtain policy equal to median voters bliss point <math>y_m</math>, so we obtain unique outcome <math>z=(x_m, y_m)</math>. Notice, that with this sequential voting, we do not get the median policy, but the median policy by dimension.