Post Mortem: 9am, November 6
The presidential vote equation was quite accurate this time. It looks like Harris will get slightly less than half of the two-party vote, maybe 49.5, which is what the equation predicted below on October 30, 2024, using the actual economic data. So the prediction error is roughly zero. Over the entire two-year period of the predictions, they have indicated a very close election regarding vote share, which turned out to be true. They have been particularly accurate in the past year as the economic forecasts got more accurate.Given the roughly zero error, it looks like there is no Trump residual as discussed below. It could be, of course, that the Trump residual is in fact, say, minus 4, but Harris's is also minus 4, resulting in a zero overall error, but none of this can be tested, so probably best to forget anything about a Trump residual.
Background:
The paper,Presidential and Congressional Vote-Share Equations: November 2022 Update,
discusses the November 2022 update of the vote-share equations. Forecasts for 2024 are also made. This discussion assumes that the paper,
Presidential and Congressional Vote-Share Equations, American Journal of Political Science, January 2009, 55-72,
has been read.
Compute your own prediction:
The following link allows you to compute your own predictions of the 2024 presidential and House elections.Compute your own predictions for 2024: Biden Not Running
Predictions:
The predictions of VP and VC for the 2024 election using the economic forecasts from the US model are:
Biden Running
NOT UPDATED
G P Z VP VC
April 25, 2024 2.23 4.62 4 51.72 49.91
January 25, 2024 1.80 4.55 4 51.46 49.77
October 26, 2023 1.40 4.78 4 51.03 49.55
July 27, 2023 1.50 5.08 3 50.06 49.04
April 30, 2023 1.20 5.17 3 49.79 48.90
January 27, 2023 1.40 4.95 3 50.07 49.04
November 23, 2022 1.70 4.83 3 50.35 49.19
Biden Not Running
G P Z VP VC
October 30, 2024 1.97 4.54 4 49.47 47.36
July 25, 2024 1.70 4.54 4 49.28 47.26
April 25, 2024 2.23 4.62 4 49.61 47.43
January 25, 2024 1.80 4.55 4 49.35 47.29
October 26, 2023 1.40 4.78 4 48.92 47.07
July 27, 2023 1.50 5.08 3 47.95 46.56
April 30, 2023 1.20 5.17 3 47.68 46.42
January 27, 2023 1.40 4.95 3 47.96 46.56
November 23, 2022 1.70 4.83 3 48.24 46.71
VP is the Democratic share of the two-party presidential vote, and VC is the
Democratic share of the two-party House vote. Click the above
"Compute your own predictions for 2024" for the definitions of G, P, and Z.
October 30, 2024, comment:
All the economic data needed for the vote predictions are available as of today. No forecasts from the US model are needed. This is my final "ex ante" prediction of the vote shares. The final prediction of the Democratic vote share is 49.47 for president and 47.36 for the House. These predictions are still close to those for November 23, 2022, and so the discussion below under November 23, 2022, is still relevant.The bottom line is that the election is predicted to be very close. The basic story has not changed much over the two years of the predictions, as you can see from the above list. Given that the estimated standard error is about 3 percentage points, the election has always been too close to call.
With Biden not running there is no incumbency advantage. The incumbency coefficient (for DPER) is 2.11, so the Democratic prediction is 2.11 percentage points lower than it otherwise would be, other things being equal. The discussion under November 23, 2022, assumed Biden was running, so the relevant numbers must be adjusted by 2.11 with Biden not running. For VC the adjustment is 2.48.
Is There a Negative Residual for Trump?
The ex post errors for VP for 2016 and 2020 are -5.4 and -3.4 respectively--- see Table 3, page 6, in Presidential and Congressional Vote-Share Equations: November 2022 Update. The negative errors mean that the equation predicted that Trump should have done better than he did. An interesting question is whether the error will again be negative. If, say, the error is -4.0 percent, then the above prediction for the Democrats rises from 49.47 to 53.47, a likely Deomcratic victory in the electoral college. There is no way to test this because of the small sample, but it is a possibility. It could be, of course, that the errors are due to above average positive characteristics of H. Clinton and Biden in the two elections.July 25, 2024, comment:
The economic forecasts from the US model changed only modestly between April 2024 and July 2024. Both the growth rate and the inflation rate are slightly lower, one negative for the Democrats and one positive. The net effect is for the Democratic vote share to fall slightly for both the president and the House. These predictions are still close to those for November 2022, and so the discussion below under November 23, 2022, is still relevant.We now know that Biden is not running, which means there is no incumbency advantage. The incumbency coefficient (for DPER) is 2.11, so the Democratic prediction is now 2.11 percentage points lower, other things being equal. The discussion below assumed Biden was running, so the relevant numbers must be adjusted by 2.11 with Biden not running. For VC the adjustment is 2.48.
April 25, 2024, comment:
The economic forecasts from the US model changed only modestly between January 2024 and April 2024. Both the growth rate and the inflation rate are slightly higher, one positive for the Democrats and one negative. The net effect is for the Democratic vote share to rise slightly for both the president and the House. These predictions are still close to those for November 2022, and so the discussion below under November 23, 2022, is still relevant.January 25, 2024, comment:
The economic forecasts from the US model changed very little between October 2023 and January 2024. The growth rate is slightly higher, and the inflation rate is slightly lower. Both of these changes increase the Democratic vote share slightly. These predictions are still close to those for November 2022, and so the discussion below under November 23, 2022, is still relevant.October 26, 2023, comment:
The economic forecasts from the US model changed very little between July 2023 and October 2023. The growth rate is slightly lower, as is the inflation rate. There is, however, one more strong growth quarter, namely 2023.3. The net effect is that the prediction for the Democrats rose by about a percentage point for president and by about a half a percentage point for the House. These predictions, however, are still close to those for November 2022, and so the discussion below under November 23, 2022, is still relevant.July 27, 2023, comment:
The economic forecasts from the US model changed very little between April 2023 and July 2023. The growth rate is slightly higher, and the inflation rate is slightly lower. Both of these help the Democrats, but only slightly. The predictions are virtually unchanged from the past predictions. Given that they are little changed from November 2022, the discussion below under November 23, 2022, is still relevant.April 30, 2023, comment:
The economic forecasts from the US model changed very little between January 2023 and April 2023. The growth rate is slightly lower and inflation is slightly higher, so the predictions for the Democrats are slightly lower. But the differences are very small, and the discussion below is still relevant.January 27, 2023, comment:
The economic forecasts from the US model changed very little between November 2022 and January 2023. The growth rate is slightly lower and inflation is slightly higher, so the predictions for the Democrats are slightly lower. But the differences are very small, and the discussion below is still relevant.November 23, 2022, comment:
The values of the non economic variables for 2024 are I = 1 (the Democrats are in power), DPER = 1 (assuming Biden runs), DUR = 0.00 (the Democrats have not been in power for two or more consecutive terms), WAR = 0, and VCC = 48.4. This latter variable is needed for the 2024 on-term House vote. Using these values, the equations for VP and VC can be written for 2024 as:VP = 49.48 + 0.708*G - 0.606*P + 0.865*Z
VC = 48.74 + 0.371*G - 0.318*P + 0.453*Z
The constant terms incorporate the non economic values just mentioned. Remember that these assume that Biden is running again. The terms will not change unless Biden does not run again.
(If Biden does not run, the constant term in the VP equation falls to 47.37 and the constant term in the VC equation falls to 46.26. This is because the coefficient estimate for DPER in the VP equation is 2.11 and the coefficient estimate for DPER in the VC equation is 2.48. According to the equations there is an advantage for the party in power if the incombent president is running again.)
Given forecasts of the three economic variables, predictions of the vote shares can be made. Using actual values through 2022.3 and forecast values from the US model on this site for 2022.4--2024.3, the values for G, P, and Z are 1.70, 4.83, and 3, respectively. These give a predicted value of 50.35 for VP and 49.19 for VC. In short, roughly a 50/50 prediction for both the presidency and the House.
To give an idea of the sensitivity of the presidential predictions, if Biden does not run this subtracts 2.13 percent from the Democratic vote share. For every one percentage point change in G, the growth rate of real per capita GDP in the first three quarters of 2024 at an annual rate, the Democratic vote share changes by about 0.7 percentage points. For every one percentage point change in P, the growth rate of the GDP deflator between 2021.1 and 2024.3 at an annual rate, the Democratic vote share changes by about 0.6 percentage points in the opposite direction. For a change in Z of 1, the Democratic vote share changes by about 0.9 percentage points. You can use the site to make your own prediction of the vote share for different values of the economic variables.