MethodLogical is now at

MethodLogical is now at

Due to some persistent technical issues we've been having with Blogger, we're now posting at Please update your RSS feeds, etc. For the time being, our new posts will automatically be mirrored here, but you'll have to visit the new site to comment.

Monday, January 24, 2011

I've Said It Before and I'll Say It Again: Democracy Doesn't Work

Steve Levitt recently raised the ire of Andrew Gelman, among others, by insinuating that it's not rational to vote. The basic argument, which Levitt really doesn't lay out too well, is that if voters acted in their narrow (monetary) self-interest they wouldn't vote, because the likelihood of altering the results of an election is incredibly low. Gelman's responses is that voting can be completely rational if you place some value on benefits to other people, that is if you're altruistic. If that's the case then the value of changing the results of an election is astronomical, so even if the chances are low it's a good bet. This story doesn't sit right with me: ask people why they vote and very few can quantify their odds of altering an election outcome. That just doesn't seem to be how people think about the problem. My friend Dan Hirschman pointed out that Gelman's argument is really more of an "existence proof" - it shows that it is possible for voting to be rational, for certain preferences, and it's valuable even if it's not the true reason people vote.
But the real reason people vote is really important, because it helps us understand what they vote for. The "rational voter" models that Levitt was implicitly attacking tend to assume that voters, or groups of voters, are voting for whatever benefits them most personally. That certainly isn't the case: what would benefit people most personally (that is, ignoring altruism) is to not vote in the first place. People are necessarily thinking about other factors than their own narrow self-interest. It's possible this means looking at benefits to others; a simpler story would be that people vote because they find it enjoyable (Gelman said this applies to himself in a response to one of my comments) or out of a sense of civic duty. Whatever the real story, the implications of this for election results are profound, and tend to be somewhat depressing; Bryan Caplan goes through them in detail in The Myth of the Rational Voter, which I can't recommend highly enough for people interested in policy and voting. To take one stark example from the book, a strong majority of Americans realize they personally benefit from trade liberalization, so rational voter models would predict support for freer trade. But many of those same people believe that trade is bad for society as a whole, and so they vote against it.
The upshot is that democracies, even well-designed ones, pick policies that are systematically worse than we might otherwise hope. And even if voters are rational and self-interested, getting democracy to work well takes a lot of work. As I pointed out in a comment on Jason Hopper's post last week, I think an emphasis on democracy as a development goal doesn't do people a whole lot of good. Part of this is practical: we tend to think of democracy as simply "holding votes and the majority rules" but in fact you need a lot of physical and institutional infrastructure to make democracy actually work; furthermore, as Hopper pointed out there are a lot of variations on democratic systems and I think which one works is very context-specific. But it's also a basic feature of democracy itself: horrible policies can be extremely popular. It's easy to think of extreme examples - Robert Mugabe was legitimately elected many times even as he devastated Zimbabwe. I'm aware that this sounds like I'm saying that we should ignore people's desires, but really I'm arguing for the opposite. Democracy doesn't do a very good job at listening to what people want and need. It's a valuable tool - it does do a good job of preventing the worst abuses of bad governments - but when we're talking about what to focus on in development I think it should be closer to #20 on the priority list than #1.


  1. I think it's important to note that the voting problem applies to ANY collective action problem. If you are pessimistic about the rationality of voting, you should be about recycling, making society more fair, etc. And so if you are serious about

    That said, I do think there is a big problem with voting. The "existence" proof above assumes a huge payoff, because for the most part we are talking about ridiculously tiny probabilities of one vote mattering (especially if you take into account the fact that close elections are decided by courts, not votes). The reason we have to assume the utility of mattering to be so high is because voting also imposes a significant cost on us. Registering is a pain in the ass, you have to mail things in or stand in lines. It's pretty much worse than the Holocaust. To justify the rationality of voting, I strongly believe we need to make it trivially easy. With 0 cost to someone, it will be worth doing even with a more moderate payoff for one vote mattering. Ben D used to always make the voting-is-irrational argument which I would counter with the Ben-is-lazy argument. He's right, though, the cost doesn't match the payoff.

    Regarding democracy, I haven't read the article you're referencing but agree whole heartedly that it is quite silly to brandish it as God's gift to the world... as if it's one thing. I'm not even clear on what we're promoting. There is Western democracy, which is something that evolved over centuries and includes highly non-democratic pieces (e.g. look at how democratic monetary policy is here). And there is the abstract idea that people should collectively be in charge. These are so far from being the same thing, and yet I feel as if they are constantly conflated.

  2. >If you are pessimistic about the rationality of voting, you should be about recycling

    I couldn't be much more pessimistic about recycling. Leaving aside the fact that lots of recycling efforts (e.g. paper) are counterproductive even if we all participate, it's just not something that's going to work by individual action. Lots of people who care about the environment do things like recycling as a sort of religious observance. I'm not a fan of that - we need a policy in place to solve the collective action problem, not guilt and ineffective changes in personal habits.

    >The reason we have to assume the utility of mattering to be so high is because voting also imposes a significant cost on us

    That's only true if we think people vote in order to change the outcome of elections. I think people value voting more directly: they feel civic pride, they've done their duty, they avoid being judged by their peers, etc.

  3. > That's only true if we think people vote in order to change the outcome of elections. I think people value voting more directly: they feel civic pride, they've done their duty, they avoid being judged by their peers, etc.

    I don't disagree with you, but insofar as that's true, it's clearly irrational :) Obviously feeling civic pride, etc. makes sense as an instrumental value given the intrinsic value of possibly mattering. But if mattering is off the table, I think one would be hard pressed to give a rational justification for caring about voting. That said, I guess you could say that given others' irrationality on the subject, voting might make sense. Like, if I say I vote because it's important to my mom that I vote, that seems like a valid reason to me, even if my mom is irrational in thinking voting matters. However--and maybe this is insufferably naive--I would hope we could converge on the rational action through discourse, and this is why I think it matters to have an actual rational foundation for voting.

  4. "Rational" is often conflated with "maximizing expected cash income" or something similar but that's not what it means. It can be perfectly rational to have civic pride in your utility function, or the opinions of others: all that's needed is that voters have complete and transitive preferences over the things they care about:

    In fact, preferences that value civic pride or the income of others or social stigma can even be consistent with the stricter requirements of expected utility maximization, which is what voter choice models tend to use.

  5. Have you ever noticed the similarities between sports and politics? I would argue that much the same energy that compels a Lions' fan (Ha!) to cheer for his team, drives a citizen to vote for his party.

    Often, it seems to me, that party attachment is detached from ideology -- and even more often it is detached from self-interest. Also, note the language used to describe elections. Words such as "fight", "win", etc. are frequently used.


Note: Only a member of this blog may post a comment.