Should We Make Everyone Vote?

Dylan Matthews has a well-meaning but ultimately misguided column in today’s Crimson arguing for compulsory voting.
Let’s start with what Dylan gets right. He is absolutely right about this: “One reason why higher economic classes’ interests are so overrepresented in government is that rich people vote at disproportionately high rates, and poor people vote at disproportionately low rates.” He is also right about this: “[L]ower voting rates among the poor aren’t just a result of choice.” There’s obviously a certain conception of choice here which libertarians and the like wouldn’t support. But all I take this to mean is that there are costs to voting which the poor bear more heavily than the rich. And because, as the libertarians well know, people respond to costs and benefits, the poor vote less frequently than the rich.
So Dylan wants to equalize those costs. He wants everyone to face the same cost-benefit analysis when they decide whether to vote, so that they truly can be said to have made that choice or not. One way to equalize costs would be to require everyone to vote, and levy some small punishment for failure to vote.
But, as Dylan notes, proposals to require voting are likely to get nowhere. Not only would Republicans reject the idea as a partisan scheme, but even non-partisans and Democrats are likely to be a little worried. If you force everyone to vote, aren’t you going to get a lot of uninformed voters? (Or more than there already are?)
If we can’t actually get a compulsory voting law, what can we do? The traditional answer is that we should lower the costs of going to the polls. Dylan gives a summary of this line of thought: “Voting is expensive. It involves taking time off work, which, if one’s employer isn’t flexible enough to allow paid voting breaks, lower-income people may not be able to afford missing. Voting also involves transportation costs, which, while trivial for wealthier individuals, impose a real cost for others.”
But this analysis misses what political scientists have found to be important, if not the most important, costs associated with voting: cognitive costs. As this paper by MIT professor Adam Berinsky shows, electoral reforms which are aimed at lowering the costs of casting a ballot (e.g. absentee balloting, vote-by-mail, early voting, and Internet voting) increase the socioeconomic bias of the electorate. That’s because most voters are occasional voters: They vote, say, only in presidential elections, and maybe some midterms if they remember. These electoral reforms make it easier for those voters to remember to vote and to actually do it. But they don’t seem to affect habitual non-voters; they don’t bring more people into the voting population. Habitual non-voters tend to face cognitive costs associated with voting, which is just to say, they don’t vote because they aren’t engaged with politics and don’t know much about it. And that group is disproportionately poor, because wealth and education are directly related to interest in and knowledge about politics. If the fine for failing to vote is low enough, these people might rationally decide to pay it rather than be drawn into the political world, about which they know and care little.
Of course, equalizing wealth and educational opportunities so that political engagement is more evenly distributed could be just as politically difficult as passing a compulsory voting law. But there are other things we can do. In Get Out the Vote, political scientists Alan Gerber and Donald Green summarize a series of experiments and conclude that the most cost-effective GOTV tactic is door-to-door canvassing. Getting one additional voter who would otherwise not have voted costs about $30 via door-to-door canvassing. So a $3 million investment could net an additional 100,000 voters. Of course, if we’re concerned about the socioeconomic makeup of the voting population, then we’ll need to target this investment effectively. I suspect that many GOTV operations are geared towards populations which are disproportionately likely to vote.
This is all very speculative and rough, of course, but the point is that all is not lost. We don’t need compulsory voting, which we aren’t likely to get, in order to increase the socioeconomic representativeness of the electorate.