Thursday, July 29, 2010

Q1: What Happened in 2000?

Mercer asks:
If economic growth is the most important factor in winning elections why did Gore barely win the popular vote? 
Great question!  And by great, I mean I like it because I've had the same question, and because, even better, there's an answer.   That's because there's a paper (pdf; I think it's the same version that was published in PS in 2001) by Larry Bartels and John Zaller (recently, by the way, mentioned by Ezra Klein) that finds that:
Gore's advantage with respect to the "fundamentals" was modest at best -- and that the election outcome was well within the range one would expect if both candidates ran more or less equally competent campaigns.
Surprised?  The key is the way that economic growth affects elections.  For better or worse, what political scientists have found is that voters have very, very short memories; the models that work best only look at election-year economic factors.  So Gore apparently got little credit for the boom years.  Moreover, what seems to matter isn't unemployment, or general economic growth, but changes in real disposable income.  And as it happened, GDP growth continued through 2000, but growth in real disposable income stalled that year, heading into the 2001 recession. 

The economy isn't the only thing that matters, as far as "fundamentals" are concerned.  Peace is better than war, and that presumably helped Gore a bit.  Ideological extremism hurts, but that wasn't really a factor in 2000 (Bartels and Zaller estimate that Bush ran a bit closer to the median voter than did Gore, but even if you disagree, it's hard to argue that either of them was perceived as far out of the mainstream).  Holding the presidency also is a negative which increases over time; they estimate that it costs about half a percentage point per term, or a 1% push towards Bush.

So, how well do Bartels and Zaller think that the fundamentals explain 2000?  They use a weighted average of 48 different models (see the paper if you want more detail), and that average misses the actual election results by half a percentage point.  In other words, the fundamentals would have predicted a very close election, and we saw a very close election. 


  1. Economic growth, per se, is not the predictor. The stock market as a barometer of social mood is actually pretty good.

    The S&P 500 gyrated between 1350 and a little over 1500 most of cy 2000, with no clear trend, consistent with a close election.

    After labor day, it fell from over 1500 to under 1300 by year's end, starting the steep decline that culminated in the 2002 double bottom.

    I'd say that for 2000 this is close, but leans against the incumbent party.


  2. JzB,

    I don't have the time for a full search, but my memory is that the stock market doesn't really seem to be a variable that adds anything once the other economic variables are accounted for. Remember, all the economic variables (well, except inflation) generally tend to move together, so it's hard to sort out the independent effects...but the people who run this stuff, again if I recall correctly, haven't found any separate stock market effect.

  3. Here's my question: Why isn't that Bartels/Zaller paper considered lame curve-fitting bullshit?

    This is actually a serious question. The best performing model for presidential elections is Hibbs. According that model, Gore seriously underperformed. When I read the Bartels/Zaller paper, I get the very strong feeling that they had a conclusion in search of regression equation. I happen to agree with all the straw men they dismiss in their argument, but is what they did really considered acceptable work in poli sci?


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

Who links to my website?