Pareto violations of parliamentary voting systems (Q2464017): Difference between revisions

The Pareto principle roughly states that a social state, an allocation or alternative should not be chosen by a society, if every member of the society prefers some other social state. The authors of this paper consider a finite set of alternatives and study four binary voting rules and evaluate how often they violate the Pareto criterion. For the first two rules, alternatives are ranked according to a predetermined order, say \(a_1 a_2 a_3 a_4\) in case of four alternatives. One is the amendment procedure, used in Anglo-American Parliaments. Alternatives are considered ``two-by-two'': the first ballot is taken between \(a_1\) and \(a_2\) in a pairwise majority contest, the winner is taken against \(a_3\), and finally, the winner is considered against \(a_4\). The other one is the successive elimination rule used in most countries of Western Europe. Alternatives are considered ``one-by-one'': the first vote is on alternative \(a_1\); if \(a_1\) wins a majority, the procedure terminates; if \(a_1\) is beaten, there is a second vote on \(a_2\), and so on, until there is a winner or a unique alternative left. Two other rules are often used in committees. Consider four alternatives, \(a_1\), \(a_2\), \(a_3\) and \(a_4\). Every voter expresses a ranking of all alternatives, but both voting procedures operate by distinguishing two subsets, say \(A_1 = \{a_1, a_3\}\) and \(A_2 = \{a_2, a_4\}\). We now distinguish the two following procedures. In the parallel elimination procedure, one subgroup of committee members, say \(S_1\), votes between \(a_1\) and \(a_3\), while the complimentary subgroup, say \(S_2\), votes between \(a_2\) and \(a_4\). Finally, the whole committee votes to elect the winning alternative, between the winners in \(A_1\) and in \(A_2\). In the split procedure, the entire committee first votes between \(A_1\) and \(A_2\), and then votes between the two alternatives of the winning subset. The authors compute the ratio (number of situations at which the Pareto criterion is violated) / (total number of situations), assuming Impartial Anonymous Culture (IAC), i.e., assuming that all situations have the same probability of occurrence, and not distinguishing between two preference profiles which only differ by the fact that the preferences of some individuals have been permuted. The authors show that the four procedures under study are not very likely to select a Pareto dominated alternative. Moreover, when the number of voters increases to infinity, the ratios tend to zero or to very small values. They also explain why in real parliaments and committees political or sociological reasons cause that these procedures are rather so Pareto stable.