President and House 2020

Post Mortem for the 2020 Elections:

The final ex ante prediction for the presidential election below was for the Democrats to receive 47.9 percent of the two-party vote. In fact Biden got 52.27 percent of the two-party vote, so the error is -4.37 percentage points. The mean absolute error (MAE) of the ex ante predictions of the 10 elections beginning with 1980 is 3.90 percentage points, as shown in Table 5 in Presidential and Congressional Vote-Share Equations: November 2018 Update. The current error is thus slightly larger in absolute value than the MAE of the past 10 elections. The ex post error will, of course, be somewhat different as the equation is reestimated through the 2020 election and the lastest revised economic data are used.

The 4.37 percentage point error is probably not large enough to warrant any specification changes when the equation is updated through the 2020 election. As was the case for the error for the 2016 election, some of overprediction of the vote for Trump may be due to his personality, but this is not possible to test.

The house vote equation did better. The last ex ante prediction was for the House Democrats to get 54.2 percent of the two-party vote, and they in fact got 51.57 percent, for an error of 2.63 percentage points. The MAE of the ex ante predictions of the last 3 on-term house elections in Table 5 is 3.50, so the error is slightly smaller than the MAE of the past 3 elections. Again, the error is probably not large enough to warrant any specification changes when the equation is updated.

Background:

The paper,
Presidential and Congressional Vote-Share Equations: November 2018 Update,
discusses the November 2018 update of the vote-share equations. Forecasts for 2020 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 2020 presidential and House elections.
Compute your own predictions for 2020

Predictions:

The predictions of VP and VC for the 2020 election using the economic forecasts from the US model were:
                             G     P      Z      VP     VC 
November 14, 2018           1.2   2.4     1     45.7   53.0 
February 28, 2019           1.2   2.2     1     45.6   52.9 
April 26, 2019              1.2   2.0     1     45.4   52.8 
July 26, 2019               1.5   2.3     0     46.2   53.3 
October 30, 2019            1.9   2.2     0     45.9   53.1 
January 30, 2020            2.2   2.0     0     45.6   52.9 
April 29, 2020               ?   ~1.9     ?      ?      ?   
July 30, 2020                ?   ~1.7   2 or 3   ?      ?   
October 29, 2020 (actual)  -5.07  1.80    3     47.9   54.2 
VP is the Democratic share of the two-party presidential vote, and VC is the Democratic share of the two-party vote in the House. Click the above "Compute your own predictions for 2020" for the definitions of G, P, and Z. The following are comments that were made at the date specified.

October 29, 2020, comment:

The preliminary third quarter NIPA data were released today. The growth rate of real per capita GDP in the third quarter at an annual rate was 32.3 percent. For the second quarter it was -31.7 percent, and for the first quarter it was -5.4 percent. The growth rate in the first three quarters at an annual rate, G in the model, was -5.07 percent. The value of the inflation rate, P, was 1.80 percent. The number of strong growth quarters, Z, was 3. These economic values lead to a prediction of VP of 47.9 and of VC of 54.2. The prediction of VP is within one standard error of 50.0, so the election is predicted to be close.

According to the NIPA data the U.S. economy came roaring back in the third quarter, which benefits the Republicans.

July 30, 2020, comment:

As in April, I did not make an economic forecast with my US model. Again, I am letting the user decide what values of G and P and Z to use. Regarding P, its value through the second quarter of 2020 is 1.7 percent, so using 1.7 will likely be fairly close to the actual.

Regarding G, the per capita growth rate at an annual rate was -5.4 percent in the first quarter of 2020 and -33.2 in the second quarter. Therefore, all that needs to be guessed is the growth rate in the third quarter. If your guess is, say, +5.0 percent, then G is (1-.054)(1-.332)(1+.05) raised to the 1/3 power and then subtract 1, which is -12.8 percent. If you use the link "Compute your own prediction," this is done for you, where you need to give as input the per capita growth rate at an annual rate for the third quarter (like 5.0). Note when choosing your estimate of the per capita growth rate that population is growing at about 0.5 percent per year at an annual rate. So subtract 0.5 from any estimate of a non per capita growth rate.

Regarding Z, the NIPA data released on July 30, 2020, were revised back to 2015. There are now two good news quarters, whereas in the earlier data there were none. The per capita growth rate at an annual rate was 3.26 percent in 2017:4 and 3.31 percent in 2018:1. Z is thus 2 or 3 depending on what one assumes about the growth rate in third quarter of 2020. If it is larger than 3.2, then Z is 3; otherwise 2.

As an example, if one uses +5.0 percent for the third quarter growth rate, which means G is -12.8 percent and Z is 3, the predicted value is 53.0 percent for VP and 57.0 percent for VC, Democratic victories.

April 29, 2020, comment:

I did not make an economic forecast with my US model this time because the model has nothing to say about the effects of pandemics. I could try to subjectively constant adjust the estimated equations, but this would only be guessing. So I am going to let the user decide what values of G and P to use. The per capita growth rate in the first quarter of 2020 was -5.2 percent at an annual rate. If your guess (annual rates) is -10 percent in the second quarter and -5 percent in the third quarter, then G is (1-.052)(1-.10)(1-.05) raised to the 1/3 power and then subtract 1, which is -6.76 percent. If you use the link "Compute your own prediction," this is done for you, where you need to give as inputs the per capita growth rates at annual rates for the second and third quarters. The value of P through the first quarter of 2020 is 1.9 percent, so using 1.9 will likely be fairly close to the actual. Also note when choosing your estimates of per capita growth rates that population is growing at about 0.5 percent per year at an annual rate. So subtract 0.5 from estimates of non per capita growth rates.

January 30, 2020, comment:

The predicted value for G from the US model is now 2.2 versus 2.0 in October, and the predicted value of P is now 2.0 versus 2.2 in October. Both of these decrease the vote share for the Democrats slightly. But the message is the same as always. See the comment dated November 14, 2018, below.

October 30, 2019, comment:

The predicted value for G from the US model is now 1.9 versus 1.5 in July, which decreases the vote share for the Democrats slightly. But the message is the same as always. See the comment dated November 14, 2018, below.

July 26, 2019, comment:

The data from the National Income and Product accounts were revised back to 2014, and the new data now show no strong growth quarters during the Trump administration. Before, 2018:2 was a strong growth quarter. The current forecast from the US model has no strong growth quarters before the election, so Z is now zero rather than one. The forecasts for G and P are both slightly higher. The net effect is that VP has risen from 45.4 to 46.2. The coefficient on Z in the VP equation is -0.792, and so a decrease in Z of 1 adds 0.792 to the Democratic vote share. VC has risen from 52.8 to 53.3. The main message about the election, however, is still the same. See the comment dated November 14, 2018, below for more details.

April 26, 2019, comment:

The economic forecasts of G, P, and Z from the US model are little changed from the previous two forecasts, so the predictions of VP and VC are little changed. See the comment dated November 14, 2018, below for more details.

February 28, 2019, comment:

The economic forecasts of G, P, and Z from the US model are little changed from the November 14, 2018, forecast, so the predictions of VP and VC are little changed. See the comment below for more details.

November 14, 2018, comment:

The economic forecasts from the US model dated October 27, 2018 are used for this prediction. The US model is forecasting modest growth for the rest of the Trump administration. The per capita growth rate (at an annual rate) in the first three quarters of 2020 (G) is forecast to be 1.20 percent. The inflation rate at an annual rate in the first 15 quarters of the Trump administration (P) is forecast to be 2.40 percent. The number of strong growth quarters is forecast to be only 1, which was the second quarter of 2018. This economic forecast is thus neither boom nor bust. Conditional on these forecasts, the prediction for VP is 45.7 and the prediction for VC is 53.0. In this situation the Democrats are predicted to lose the presidential election by 4.3 percent points, larger than one standard error, and win the two-party House vote by 3.0 percent points, a little over one standard error. See the discussion of Table 6 on page 12 in the first paper linked above for alternative predictions, one with a booming economy and one with a recession.

Why is the presidential equation so pessimistic for the Democrats? The current case is the best possible one for the Republicans according to the equation: President running again and no negative duration effect. In this case it takes a weak economy to have the voting equation predict the Democrats getting close to 50 percent of the two-party vote. This analysis, of course, does not take into account the personalities of the candidates.