Presidential Vote Equation--October 27, 2000
The NIPA data for the third quarter of 2000 were announced today, and so actual values of all the economic variables are available for the vote prediction. There are 8 quarters from 1997:1 on in which the per capita real growth rate has exceeded 3.2 percent: 1997:1, 1997:2, 1998:1, 1998:4, 1999:3, 1999:4, 2000:1, and 2000:2. There are thus 8 good news quarters that are relevant for the 2000 election. The per capita growth rate in the first three quarters of 2000 (at an annual rate) is 3.5 percent, and the inflation rate in the past 15 quarters (at an annual rate) is 1.7 percent. The economic variables are thus:
  • n = 8
  • g3 = 3.5
  • p15 = 1.7
The predicted Democratic share of the two party vote (V) is thus .508:
  • .508 = .423 + .0070*3.5 - .0072*1.7 + .0091*8
The Democrats are thus predicted to win with 50.8 percent of the two party vote. Given that the standard error of the equation is 2.15 percentage points, the election is essentially predicted to be too close to call. The equation will have done well if the election is close regardless of who wins. If either Bush or Gore wins by a fairly wide margin, say with 54 or 55 percent of the two party vote, the equation will have done poorly.

Some people ask with the economy as good as it is why isn't the equation predicting a bigger Democratic victory. The duration variable is working against the Democrats, and unlike when Clinton ran in 1996, there is no person advantage for Gore since he is not the incumbent President running again. In addition, there is a slight historical bias in favor of the Republicans, other things being equal. The Democrats thus start off behind, and it takes a strong economy for them to win. The economy is predicted to be fairly strong, but it is by no means the strongest in history regarding the variables n and g3 (see the data in Table A-1 at the end of Fair (1998)). According to the equation the economy is just strong enough to have the Democrats squeak by, but little confidence can be placed on this prediction given the size of the standard error.