Information aggregation with runoff voting (Q1995290)

This paper investigates the information aggregation properties of runoff rules in general informational environments with an arbitrary number of alternatives. A multiple runoff rule is defined, in which voting takes place in multiple rounds, and the alternative with the least number of votes is eliminated in each round, until a unique winner is obtained. Voters are divided into common value voters, also referred to as truth-seeking, and diverse instances of partisan voters. While common value voters wish to efficiently aggregate their pieces of possibly conflicting information to make a correct decision and may disagree on which alternative is the best match due to private information, partisan voters support certain candidates by their own private reasons. It is then shown that, under general assumptions regarding the information environment and partisans' preferences, multi-round runoff voting achieves asymptotically full-information equivalence. The same investigation approach is applied to the plurality rule and the two-round runoff system. It is established that when there are more than two alternatives, the plurality rule does not always admit an efficient equilibrium, while the two-round runoff system can aggregate information efficiently as long as there are at most four alternatives.