« October Baseball again | Main | Up where the air is clear »

Fundamentals, Fluctuations and Feels

So. The election.

Your Humble Blogger has over the last couple of decades, been convinced that the fundamentals view of general Presidential campaigns is closer to accurate than the version that takes the candidates and their campaigns to be dispositive. The fundamentals are defined and weighted differently, but are generally some combination of the current president’s popularity, the state of the economy (usually its growth, measured over some period before the election), the time the current party has held the presidency and (for some models) the quantity of casualties in foreign wars. The actual candidates and campaigns sway a few percent of the vote, in this view, but the bulk of them are either partisans (either officially or practically) who will eventually vote with their Party or low-information voters who will mostly vote on how they feel about the incumbent President, whether they know it or not.

Now, I have to admit that I find this version persuasive because I find it appealing. It says that for an unscrupulous and ambitious rogue, the path to re-election is to avoid personal scandals that drag down personal popularity, avoid risky military adventuring, and implement policies that encourage widespread prosperity. In other words, govern well. In turn that their advisors, looking for jobs in a new administration or running for the post themselves, have the highest incentive to govern well and to pressure the President to govern well, that incentive being their ruthless and unscrupulous ambition. That’s how James Madison set up the system to work, that’s how I want the system to work, and so I want to believe that’s how the system in fact does work.

It’s not that simple, of course—for one thing, we saw in 2004 that an incumbent President can be fairly popular due to things other than competent governance and policy success. And I believe that Ronald Reagan’s re-election in 1984 was made much easier by how well people thought he and the Democratic Congress dealt with the recession that was caused by his policy failures. Still, in general, the evidence does look persuasive that almost the entire electorate’s vote can be accounted for by the basic background partisanship, the economy and the incumbent’s popularity. At the moment, Our Only President is fairly popular, and the economy, while not great, seems to be in the sort of shape that makes me reasonably confident of victory.

So, look: it’s a month until the election. I know (or at least I hope) that y’all are passionate about the outcome. I surely am. I’m terrified that Donald Trump may become President of the United States, which could totally happen. But don’t place too much meaning on daily shifts in the polling and thus in the odds of winning. Places like FiveThirtyEight or The Upshot or even RealClear Politics will have daily updates that fluctuate over the month. There is an incentive for most of those sites to use a formula that is sensitive to those fluctuations rather than smoothing them. Frequent changes result in more clicks, of course, and also I would guess most of the readers want to know the daily ups and downs. I don’t think it’s wise. I think we’d be all much better off with the smoothed-out version. Frankly, I think the up-and-down is mostly responding to noise in the data; if a few polls are outliers in the same direction one day, the graph goes wild. But even if it were accurately informing you about who had a bad day and who had a good one, it wouldn’t tell you much about who is likely to win in November. That is largely baked in already, even if we don’t know it yet.

And even that, really, is not the point. The outcome of the election is important, but it’s not the point. Electing the right person (or not electing the wrong one, as in this year) makes an enormous difference in people’s lives. But it doesn’t change the big structural problems. Even if Hillary Clinton wins the election, if my Party fails to take the Senate, she will be unable to govern effectively. I’m worried that whatever happens, the Other Party will not recover from its dire dysfunction. The election is one factor in changing incentives and disincentives. But we’re not electing a President to solve our problems. And we’re certainly not electing a President to solve our problems while we sit back and neglect to participate. The Eighth of November will be very important; the Ninth will be important, too. And December, and January of 2017, and all the days to come. The fluctuating graph is a distraction from that.

By the way, Gentle Readers might be interested in a Twitter Rant from David S. Bernstein, starting with this tweet, in which he talks about the market logic of the dissemination of conspiracy theories into mainstream conservatism. This is scary stuff independent of the outcome of the election, and in some ways the market forces will be more compelling under a Hillary Clinton administration than a Donald Trump one—the market will be more lucrative, certainly. I’m not saying it would be worse for the country (or the world), but it might make it even more difficult to include the Other Party in governance, which is also bad. Not to despair, though—I’m trying (in part by ignoring those fluctuating graphs and hoping those fundamentals come through) to keep my balance on that keel as even as possible, giving in neither to despair or elation, and keeping my eyes on the rest of the calendar before and after the ballot-shower.

Tolerabimus quod tolerare debemus,


I see what you're saying, and I think it mostly makes sense, but I have one objection to one part. You wrote:

"Places like FiveThirtyEight or The Upshot or even RealClear Politics will have daily updates that fluctuate over the month. There is an incentive for most of those sites to use a formula that is sensitive to those fluctuations rather than smoothing them."

I would say that on the contrary, FiveThirtyEight (a) has three different models that take three different approaches to fluctuations, one of which ("polls plus") does a lot of smoothing/damping, and (b) is very aware of the fundamentals approach; that's where I first encountered the idea, and their "polls plus" model is explicitly rooted in fundamentals.

Yes, FiveThirtyEight knows about the fundamentals theory and Nate Silver's disagreement with it is quite nuanced. And, I think, still wrong. And I think that site and others are publicizing the sensitive one more than the more-accurate smoothed one that most of them have, either on purpose or just because that's the one that has (illusory) news every day and that people link to. If you really have to monitor one, choose the smooth one. But the stuff that I've seen posted most often is the sensitive one, because that makes news a hell of a lot faster than the smooth one. Still, you are better off not looking at them at all, or only in passing and not focusing on them.


Post a comment

Please join in. Comments on older posts will be held for moderation. Don't be a jerk. Eat fruit.