In May, Indiana voters decided 10 school referendums. Nine passed. This November, two referendums were decided.
Since November 2009, 66 percent of referendums in May have passed. Only 38 percent of referendums in November have passed.
Let’s think up a story about voting and see if we can explain
I’m an economist, so there’s only one way for me to tell this story: Assume people act in their own self-interest. Economists call such behavior rational, which is a pretty dismal view about people’s motivations. Let’s tell a rational voter story anyway.
Rational voters weigh the benefits and costs to themselves when they decide whether to vote and how to vote. Suppose parents benefit directly from added school spending on better facilities, teachers and staff. Suppose both parents and nonparents get indirect benefits, like added economic development from better-educated employees. The costs of added property taxes are spread among all property owners, including renters, if landlords raise rents when property taxes go up.
Parents get direct and indirect benefits from school spending. Nonparents only get the indirect benefits. The tax is spread equally over all taxpayers. In our story, that means parents get positive net benefits and nonparents get negative net benefits.
So, rational parents will vote yes for added school spending, and rational nonparents will vote no.
Stop right there! We all know that plenty of parents vote no, and plenty of nonparents vote yes.
Other things than self-interested benefits and costs must matter when people decide how to vote. But let’s keep going. Maybe there are just enough rational folks to sway referendum results one way or another.
This next part matters a lot. Since parents are a minority of voters, their positive net benefits are larger per person. Negative net benefits are spread widely among nonparents, so they are smaller per person.
Let’s also suppose that voting has costs. You must find the time to go to the poll and wait in line, fish out your ID and cast your vote. You use gas. If it’s raining, you get wet.
In our story, parents vote yes, and with their large net benefits, they’re more likely to bear the costs of voting and show up at the polls. In our story, nonparents vote no, but their negative net benefits are small, so the costs of going to the polls may keep them at home. It’s a contest between the greater motivation of parents and the greater number of nonparents.
What if the referendum is not the only thing on the ballot? Suppose we’re electing a president, a governor, some federal and state representatives? That makes everyone’s benefits and costs larger, so voters with negative net benefits from school spending are more likely to show up to vote for their candidates and against the referendum.
In May, the referendum is more likely to be alone on the ballot, so it’s more likely to pass. In November, there are other things to vote about. The referendum is less likely to pass. That’s what we see
Here’s another test. In odd years, the November ballot has no federal or state candidates, so those Novembers should look like May. In the odd years since 2009, eight of 17 school referendums have passed (47 percent). In the even years since 2010, seven of 23 have passed (30 percent). That matches the story.
What else could we test? If it’s raining, the cost of going to the polls is higher. That should knock out more of the other opponents than parent-supporters. Referendums might pass more often in bad weather. We’d need to collect weather data for the 104 school referendums so far. Sounds like a job for a graduate student.
Indiana’s results look pretty good for the rational voter story. There’s a big problem, though. Rational voters vote only if they think that their vote will make the difference. It’s pretty unlikely that a referendum will be decided by that one vote. You’re more likely to have an auto accident on the way to the polls instead.
A truly rational voter wouldn’t vote at all.
Anyway, it’s a story that makes predictions. And we get to test those predictions every May and November.