Peter Levine

« the Institute of Development Studies | Main | dulce et decorum est »

July 14, 2004

too close to call

I've previously advertised this impressive article by Larry Bartels and John Zaller, which finds that the change in real personal disposable income is the best predictor of an incumbent president's share of the vote. The ethical implications of this model bother me--rational people shouldn't vote for the incumbent just because their spending power has grown during a short period in the recent past. However, I can well believe that the Bartels/Zaller model accurately predicts mass behavior.

So who will win? As I calculate it (and my math is fallible) the annual change in real disposable income during the first five months of 2004 was 2.3%. Bartels and Zaller's regression line (see their figure 1) predicts that 2.3% growth would buy the incumbent just over 50% of the popular vote. But there's a substantial margin of error (because the model is based on a small number of cases). Besides, we don't know what will happen to real disposable income between now and November. Thus we ought to expect a very close and uncertain contest, but if current economic conditions are sustained, Bush has the edge. Most other election models give him a bigger margin.

(An interesting footnote: Bartels and Zaller claim that if Clinton had spent the federal surplus in the form of a tax cut, thus raising disposable incomes, Al Gore would have been elected in 2000. Policy won over politics.)

Posted by peterlevine at July 14, 2004 12:37 PM

Comments

Post a comment