Various statistical measures indicate a high degree of accuracy for our model in projecting the results of the 19 midterm elections from 1946 to 2018. We used those results to forecast the swing in House seats for the upcoming midterm election in November.
In analyzing the election results for the 19 midterm elections between 1946 to 2018, we found that a sitting president’s political party almost always experiences a net loss in House seats. Since 1946, the president’s party has lost seats in every midterm election except in 1998 and 2002. In 1998, the Democrats, under President Bill Clinton, eked out a small gain of four House seats. Four years later, in 2002, the Republicans, under President George W. Bush, gained a scant eight seats.
Every other midterm election since 1946 resulted in losses—usually big ones—in House seats for the president’s political party. The average loss, including the small gains in 1998 and 2002, was 27 seats. The magnitude of the losses in some midterms, however, was much higher than the average. For example, there were losses of more than 50 seats for the president’s party in the midterm elections of 1946, 1958, 1994, and 2010.
Our study identified a number of factors that explain these wide variations in midterm election results. The two we found that we believe will figure most prominently in forecasting this November’s midterm House election results include the price of gasoline and the Gallup presidential approval rankings.
We were surprised to find that overall price changes as measured by the Consumer Price Index had no measurable impact on midterm House election results. Gasoline prices, however, were found to be highly significant in terms of influencing voters. Since people are reminded virtually every day about the price per gallon of gas at every gasoline station they drive by, voters are likely to be more upset about gas prices than overall rates of inflation. In addition to finding that voters blame the presidential party for rising gas prices, we found that voters credit the president’s party when gas prices fall.
During the three midterm elections that experienced the most significant declines in gas prices, the president’s party had its best showings. In fact, the only two midterm elections where the president’s party gained House seats (1998 and 2002) were during years when gas prices declined.
But when gas prices increased the most during a midterm election year (37 percent in 1974), the president’s party lost 48 seats. The second-highest year of increasing gas prices was 2010 (19 percent) when the president’s party lost 64 seats.
Over the entire period, we found that for every increase of 10 percent in gas prices, the president’s party lost, on average, 4.3 seats. This suggests that this year’s estimated annual increase in gas prices of a historical high of 61 percent will lead to a loss of 29 House seats for the Democratic Party.
Our study also found that the Gallup presidential approval rating affects House election outcomes. Specifically, we found that for every decrease of 1 percent in a president’s approval rating, the president’s party will lose an average of one seat. Hence, the fact that President Joe Biden’s current approval ranking of 41 percent is nine points below the historical average ratio of 50 points to an additional loss of nine House seats for the Democratic Party this fall.
Other variables in our forecasting model are estimated to lead to the loss of an additional 15 seats beyond the loss of 29 seats resulting from high gas prices and the loss of nine seats because of Biden’s currently low presidential approval rating. That points to a total loss in Democratic seats of 53 in November. Such a loss would shift the current House division of 222 Democratic seats and 212 Republican seats to 169 for the Democrats and 265 for the Republicans, clearly pointing to a decisive majority for the Republican Party.
If Biden is pinning his hopes on a Democratic victory in the upcoming midterm election to convince him to run for a second term in 2024, the outlook for him looks grim.
