Accountability and dishonesty
A while back, John Sides and I had a discussion, motivated by Seth Masket’s new book No Middle Ground, about the relevance of the “true preferences” of politicians. That got me thinking about the sincerity of preferences, which led to a post a few days ago, where I argued that we shouldn’t care so much if politicians offer us arguments that they themselves don’t seem to believe. The discussion got pretty metaphysical, which is cool. But the general idea is something I would hope journalists and commentators would pick up on. Having gone deep on the arguments, let me make the possibly less intense claim: We shouldn’t care so much whether politicians actually favor the policies they say they stand for (forget about why they say they favor them). To get there, we need to stop off at some political theory and some political science.
First, the political theory:
As much as we might value character and principles, we generally think that democracy is supposed to be about accountability and representation. By accountability, I mean that when political leaders do something, someone else — ultimately voters — ought to be able to identify who has done that something, and then punish or reward them if they like it. We like to think that such accountability enables representation. Whether you think representation is about delegation (your representative is supposed to do what you would do if you were there) or trusteeship (your rep is supposed to do what they think is right, and you just try to put someone into office who you think will do god things), you retrospectively get to reassess your decision. Knowing this, representatives will try to be good delegates, trustees, or whatever it is they think you want.
Then, the political science:
So far so good on accountability, but the obvious next question is, accountable to whom? We think “the voters,” and there is at least some good evidence that they can hold leaders accountable for broad successes and failures, like good economic performance and poor foreign policy. But a different group that politicians heed is political activists, notably partisan activists. After all, the party hold the keys to renomination. And party activists are the foot soldiers in re-election campaigns. Their cooperation is also critical in policy fights. So politicians are accountable to voters, but also to the party.
That’s what Masket was arguing. And what I (with co-authors) and Masket and I (again with co-authors) have argued. If we’re right, then what it means is that if you want to know what a politician will do, you should think about what that politician thinks voters and his party want him to do, more than what he himself thinks is right. (Of course, the voters and the parties would do well to put people who share their true preferences into office, but that assumes you could ever know what they really think, given their incentives to deviate).
Now for a two contemporary examples:
1. As several people have pointed out recently, Obama has the same policy preference on gay marriage as Carrie Prejean. Conservatives seem irked that liberals are mad at Prejean but not at His O-liness. They think the same thing. (And, come to think of it, Kerry and Bush thought the same thing in 2004, and Palin and Biden agreed to it in their debate as well).
But Obama (and Kerry, and Biden) are Democrats, and the Democratic Party includes within it activists for gay rights, including marriage. And so Obama opposed Proposition 8, even as he doesn’t move to revoke Don’t Ask Don’t Tell. Liberals can be mad that he’s not moving fast enough, and conservatives can suspect that his true feelings are hidden. Obama’s position is best understood as a struggle to be accountable to the Democratic activists, who favor gay marriage, and the genereal election voters, who are on balance opposed.
2. When Arlen Spector skipped out on the GOP, a lot of people called him opportunistic. He’s not doing it because he’s in principle opposed to the Republican Party, but because he wants to keep his Senate seat. But political scientists correctly asked just how much a switch from being accountable to one party to another matters, especially for someone who seems to be not that accountable to either (and yet apparently satisfactory to the voters of Pennsylvania.)
Comments
Hans:
This is a very interesting discussion. I agree completely with the idea that politicians are constrained by both parties and the broader electorate. But I think these two factors are more significant for the things a politician doesn’t care about—you might call them “position-taking” issues. Many politicians (though not all) have one or more issues that they passionately care about, that are definitely not position-taking issues, and on those topics their own point of view is likely to be far more relevant. They take the necessary positions on all the other things in order to give them the flexibility to pursue the stuff they care about more aggressively. The guy I worked for on Capitol Hill focused most of his energy on foreign policy issues, even though neither the party activists nor his constituents cared if he did or not. And he definitely had a point of view on those issues that seemed unswayed by political context. As Richard Hall has pointed out, casting a vote in Congress is no big deal, but doing the hard work to put a bill together is a big deal, and you have to care about an issue to want to go to that kind of trouble. That doesn’t mean that the sort of solutions you pursue for your chosen issue won’t be further constrained by the politics of that issue—just that your unique perspective is probably a lot more relevant in that process than it is in the act of taking positions in the media or casting votes on the floor.
Posted by: Eric McGhee | May 22, 2009 05:19 PM
As you know, I tend to agree with this perspective on parties. I also agree that it makes sense to think about politicians’ behavior as shaped by incentives and not worry about discerning their “true” preferences. But I would just like to push you on a couple points.
First, thinking about the incentives of a primary electorate, they should want to nominate the most extreme candidate who will still be elected. But if both parties’ primary electorates behave this way, this it becomes equivalent to a simple Downsian candidate positioning model and we should expect convergence to the median. In other words, based only on what you have said, it’s hard for me to imagine an equilibrium where the party’s policy demanders actually pressure candidates to take off-median positions.
Second, representation modelers often break representation problems down into two categories: moral hazard (where politicians don’t have incentives to behave in ways that serve their constituents’ interests) and adverse selection (where elected politicians have different preferences from their constituents). I think what you’re saying in this post is, if moral hazard is eliminated, good representation is assured, regardless of any adverse selection. The logic of that seems intuitively plausible, but I wonder if anyone has tried to prove it formally. Does anyone know if a game theorist has already proven this proposition correct or incorrect?
Posted by: Jon Ladd | May 22, 2009 11:37 PM
@Eric: That’s a great point. Whatever else they are, politicians are people who cared enough about politics that they made it their vocation. It’s hard to believe they would have no policy preferences. Still, two thoughts:
1. I think what matters is not whether the issue is something the member cares about or not, but rather whether it is something the party and the voters care about. There is a lot of policy that the voters just don’t pay attention to. And there is a lot that the parties may be willing to tolerate. But these are often areas that are not of much interest to media and other commentators, which is where I am aiming my point. We tend to “care” what a politician thinks when it is in conflict with something, and when that is the case, I think our first impulse is to figure out to whom the member wants to be accountable.
2. Accountability is usually a negative power. If the voters and the party really don’t care, and never will, then our conception of representation doesn’t really know what to do with that member. On the other hand, if some downstream effect of the policies he pursues is bad, and if accountability works, then he’ll face the music when the time comes. The issue, then, is whether we can identify him as the one responsible. And to do that, we again don’t need to know so much what he wants to do, but what he accomplishes.
Nevertheless, the point is well taken that there is a lot of room for personal preferences.
Posted by: Hans Noel | May 23, 2009 04:06 PM
@Jon: Both good points. On your first, there are several papers, notably by Calvert and by Wittman, that model politicians who have preferences for policy as well as for office. The microfoundations are a little different, but the general effect is the same as a politician torn between the median voter and a non-median party ideal point. These models generally assume probabilistic voting, and so the result is not the median, because if the other party is at the median, the drop in the probability of victory from a small deviation from the median may not offset the increase in the value of getting better policy. But the upshot is no clear prediction. In Bawn, Cohen, Karol, Masket, Noel and Zaller, we claim that just about any policy that is even close to the median is not extreme enough for the voters to notice or care about, so the result is always some deviation. But there, too, the prediction is not sharp.
On your second point, that’s an interesting rephrasing. The representation context is not the same as, say, the insurance context, so I’m not sure I can think of any relevant models. But I especially like the “if” that you are pointing out. “If moral hazard is eliminated” in your terms is equivalent to saying that the accountability mechanisms we might put in place are actually working. (We might also think that accountability to a political party is a bad thing. I don’t think so, but that’s a reasonable position.) So the argument is that we should be focusing more on addressing moral hazard (accountability mechanisms) rather than worrying so much about selecting people with bad preferences.
It’s also worth noting, however, that one way that parties (and voters) can get accountability would be to elect people whose preferences are in line with them. So, in the terms you offer, if you eliminate get the selection right, you don’t have to worry so much about how the selected will behave.
Posted by: Hans Noel | May 23, 2009 04:08 PM