Strategic Voting in Two-Party Legislative Elections

It is commonly thought that in an election with two parties there can be no strategic voting - voters simply vote for their preferred candidate. In this paper, I show that strategic voting comes to the fore in legislative elections with multiple policy dimensions. In sharp contrast to single-district elections, the intensity of a voter’s preference on each dimension is irrelevant for her voting decision. Instead, she votes solely based on the dimension which is most likely to be pivotal in the legislature. Anticipating this behaviour, candidates put forward a different set of policies than they would in a single-district election. For large elections I show that the implemented policy bundle: (a) is uniquely pinned down by voter preferences, (b) is preferred by a majority of districts on each dimension, (c) is a Condorcet winner, if one exists. These properties are not guaranteed in a single-district election. Furthermore, I show that (i) parliamentary systems generate superior policies to presidential systems and (ii) voter polarisation affects outcomes in single-district elections but not legislative elections.