| Abstract: | Simple plurality election systems (commonly known as ‘firstpast the post’) are often associated with the dominanceof two political parties. Such systems tend to reward leadingparties with a disproportionately large number of seats (the‘mechanical’ effect) and provoke tactical voting,where voters switch away from trailing parties (the ‘psychological’effect). We view tactical voting as a coordination problem.A group of voters wish to prevent a win by a disliked party(such as the Conservatives in recent UK elections) and mustpartially coordinate behind a single challenger (such as Labouror the Liberal Democrats) in order to do this. Crucially, votershave limited information on the situation within their constituencyand hence there is no common knowledge of the game being played—tacticalvoting is a global game. We show that in this setting, voterswill only partially coordinate. Furthermore, tactical votingexhibits negative feedback—tactical voting by others reducesthe incentive for an individual to vote tactically, since theybecome concerned that they may switch in the wrong direction.We calibrate our model, and apply it to the UK General Electionof 1997. Throughout England, we find that the ‘mechanical’and ‘psychological’ effects tend to offset eachother: tactical voting serves to reverse the Conservative biasthat results from the geographic distribution of preferences. |
