Why does New Hampshire matter: simultaneous v.s. sequential election with multiple candidates

Abstract

I study and compare preference aggregation in a simultaneous and a sequential multicandidate election. Voters have perfect information about their own preference but do not know the median voter’s preference. A voter has an incentive to vote for her second choice for fear that a tie between her second and third choice is more likely than she would like. Therefore, a voter may want to coordinate with supports of her second choice. I show that when voters. preference intensity for their .rst choice is moderate, in the limit as the electorate increases, there is a unique equilibrium in the voting game within one voting round exhibiting multi-candidate support. In such an equilibrium, the ex ante probability that a candidate wins increases in her supporters. preference intensity and decreases in her opponents’ preference intensity. There is too much coordination with supporters of a voter’s second choice in that sometimes the median voter’s second choice wins the election. A sequential election allows later voters to coordinate with earlier voters. Therefore, in the last voting round, votes are split between the two front runners. The voting outcome in the first round affects the voting behavior of the second round. A victory of a voter’s favorite candidate in the first round may change the outcome of the second round from the voter’s second choice to her favorite candidate or from her last choice to her second choice. When preference intensity is moderate, voters vote more for their first choice if they vote first in a sequential election than in a simultaneous election, and the probability that the median voter’s first choice does not win a voting round is smaller if voting takes place sequentially. Using this model, I show that in a sequential election with ex ante identical states, no matter who the median voter in New Hampshire is, voting first is better than voting second if preference intensity is small.