I’ll be talking this coming weekend with the fine Australian libertarian community about public choice problems with voting; I’ll also be in a panel discussion on nanny state issues.
Voting’s been something of a longstanding problem in public choice. On purely instrumental accounts, where people vote to make it slightly more likely that their preferred policies are enacted, it’s hard to explain voting at existing levels of turnout. The infinitessimal probability of casting the decisive vote wipes out any potential utility difference that could attach to the different candidates’ likely policies. And so we look to intrinsic accounts of voting where people derive utility from the act of voting itself.
But even those intrinsic accounts have problems. If people vote out of civic duty, and if there are myriad things people could do instead to demonstrate their dutifulness, why vote instead of spending the same amount of time donating blood or helping at a homeless shelter? You’d get the intrinsic benefits of doing something dutiful while also potentially doing something useful. If people vote because team or group affiliation make them feel good, or if whatever motivates people to get to the polling booth doesn’t also motivate them to put time and effort into their choice, then there’s no reason to believe that there’s any connection between the median voter’s choice and desirable outcomes. We have rather a fair bit of evidence that voters do not seem to know very much.
So why do people vote? Stefano DellaVigna, John List and Ulrike Malmendier say it’s to avoid the pain of having either to lie or to admit to others that you didn’t bother voting. In a rather nice field experiment, they varied payment to survey respondents, varied the up-front estimate of the time cost of the survey and varied how much longer they said it would take voters to complete the survey relative to non-voters. As they had voting records that told them which respondents had in fact voted, they could use the design in combination with differential response and lying rates by voters and non-voters to elicit the disutility of having to talk about having voted. In a year without a Presidential election, that value was estimated at $10-$15: being asked about whether you voted, when you hadn’t voted, cost the non-voter $10-$15; the figure would presumably rise in a Presidential election as more people would ask.
The study has implications for campaigns’ get-out-the-vote efforts, which often involve enlisting people to ask individuals whether they voted, Ms. Malmendier said. “They ask friends and neighbors to ask if you plan to vote. They will ask you and then call you on Election Day… Our paper is saying we should not forget that people get a ‘disutility’ from that. They really don’t like it. Even if they’re voters, they really don’t like being asked.”
Ah, but that cost is what makes it work if the disutility comes either from saying you won’t vote or from fearing that they’ll know you didn’t. Campaigns are acting rationally and probably reasonably expect that their supporters won’t vote for the other guy out of spite after a GOTV call.
Political parties aren’t subject to the US national “do not call” registry; it’s perhaps worth re-thinking that exemption.