Buying a Lottery Ticket to Help the Poor

Abstract

A core problem of rational-actor models for politics is the seeming irrationality of voting, namely, that it is irrational for voters to vote and to be informed about politics. A simple cost-benefit analysis indicates that the probability of one's vote being decisive is so small that the costs of voting exceed the benefits. Some authors propose that civic duty or expressive voting explain why individuals vote. However, these explanations are not completely satisfactory. Five facts characterize American voting behavior: (1) the costs of voting, e.g. registration, etc. affect the probability of voting; (2) turnout varies substantially according to the type of election. In the United States, turnout for presidential elections is about 50%; for off-year elections it is about 35%; and for state and local elections it is about 25%; (3) more educated voters have a higher probability of voting; (4) individuals have a tendency to abstain from voting for races appearing lower on the ballot (the roll-off phenomenon); and (5) voters often vote strategically when their most preferred candidate has no chance of winning. Neither civic duty nor expressive voting can explain all five of these facts of voting. I argue that any successful analysis of voting must not only explain the positive turnout in elections, but also these five facts of voting. I expand the basic voting calculus model by incorporating both a pure altruistic motive and a more extensive incomplete-information component. The incorporation into the standard rational-actor model of voting of a low-grade sense of altruism is essential. The resulting analysis allows me to explain the five facts of American voting behavior.