How long is a piece of string? An exploration of multi-winner approval voting and ballot-length restrictions

Abstract: Multi-winner approval elections are seen in a variety of settings ranging from academic societies and associations to public elections. In such elections, it is often the case that ballot-length restrictions are enforced; that is, where voters have a limit on the number of candidates which they can vote for. Despite this common feature, there does not seem to be any theoretical justification for ballot-length restrictions (Laslier and Van der Straeten, 2016). This work endogenously derives the set of voter best-response ballot lengths under complete information and with general assumptions on voter utilities and voting rules. These results provide justification for some ballot-length restrictions observed in practice, however when considering equilibrium outcomes our analysis shows that this justification is no longer valid. Equilibrium analysis is considered for voters with lazy and truth-bias second-order tendencies and the equilibrium solution concept is pure-Nash equilibria. The key insights show that ballot-length restrictions or institutional features which make voting costly may lead to instability in election outcomes when voters have diverse preferences, via the non-existence of equilibria. On the other hand, when equilibria do exist they satisfy desirable properties which are not guaranteed by equilibria attained under costless voting and in the absence of ballot-length restrictions. In summary our results highlight a stark trade-off between stable and desirable election outcomes.