Five models dominate academic analyses of elections for the House of Representatives. They are: (1) the surge and decline cycle of presidential elections and the successive off-year election; (2) the presidential performance model; (3) the retirement model; (4) the campaign spending model; and (5) the candidate qual- ity model. The expected results for the 2022 Congressional election according to the first three models are presented here. Senate elections are not considered.
The models differ in their predicted result. The smallest impact on 2022 is as- sociated with the surge and decline cycle; the largest impact is found in retire- ments. The average of the three predictions for 2022 finds the Democratic Party losing 23 seats, giving the Republicans control of the House of Representatives for the 118th Congress.
Surge and Decline: The Presidential and Congressional Election Cycle
The president’s party almost loses seats in the off-year Congressional election, particularly for the first term. In the lifetime of the contemporary American par- ty system, this generalization has been inaccurate only twice: in 1934 and 2002, and for reasons covered by the other models.
Each party can be assured of success in some districts because of the partisan- ship of the voters. However, other districts will have a more equal partisan bal- ance, making them competitive and the outcome will reflect the short-term forc- es (the STFs) of the moment. Peace, prosperity, the state of the economy, the country’s foreign relationships, the misfeasance or malfeasance of current office holders, and the qualities of the competing candidates provide the STFs that tip the balance.
A major STF for competitive congressional districts is the presidential election. The better a presidential candidate does, the greater the vote will be for the Congressional candidates of that president’s party. That passing of the presi- dent’s appeal to his or her down-ballot fellows will often elect the latter even when the district has a slight tilt to the opposing party and certainly when it is competitive. The down-ballot effect is weaker than it was decades ago, but still visible and the source of surge effect in down-ballot elections in presidential years.
What the top-of-the-ticket gives is then taken away in the subsequent Congres- sional elections, with the size of the off-year decline proportional to the on-year surge: the bigger the presidential surge, the bigger the off-year decline.
The following figure plots the surge and decline relationship from 1962 through 2018. The horizontal axis is the number of seats won in the presidential elec- tion that exceed the average number of seats held by the party during the peri- od. The vertical axis is the number of seats lost in the subsequent Congression- al election. The slope in the figure indicates that average change in House seats won from the presidential to the off-year election.
It predicts Democratic seat loses in 2022. Although the Democratic seat share declined with the 2020 election, they still won 10 more seats than their average since 1994. Biden provided a small down-ballot pull. The asterisk on the slope is where the 2020 Democratic wins place them for expected 2022 outcomes. The horizontal line with the arrow, pointing to the vertical axis, is the expected 2022 outcome.
The surge in 2020, though small, predicts a 2022 loss of approximately 12-13 seats by the Democrats (note the point on the vertical axis met by the arrow).
Presidential Approval and Congressional Election Results
An immediate source of this decline is the assessment of Biden’s job perfor- mance. When a president enjoys a high job approval rating, the retreat of the electorate from the positivity that voters felt toward the winning presidential candidate is lessened. It gives his or her fellow Democrats or Republicans run- ning for a Congressional seat a campaign asset: the popularity of the president. When the approval level is low, the president’s job performance is an asset to the challenger. The overall pattern is apparent in the figure below.
Presidents viewed as not doing well lose the excess seats won during the presi- dential contest two years previously, with the loss proportional to the negativity of the president’s job approval. On the other hand, even more seats than those provided by the presidential election surge are won by the candidates of the in- cumbent president’s party when the job performance rating is strongly positive.
As of early December, with President Biden’s overall job approval in the low 40s, the model expects an 18 seat loss in the 2022 Congressional election.
Retirements as a Harbinger of Seat Losses
This retirement effect estimation – for any office - assumes that most avoid be- coming candidates if they expect to lose. Would-be challengers wait for a prom- ising election milieu since a loss is not an advantage for future efforts. Incum- bents retire since retirement is a better end to a political career than defeat. Congressional incumbents and their advisers are savvy political professionals, who anticipate when defeat is likely, making retirements an indicator of the ex- pected aggregate election outcome.
As the following figure indicates, the balance of retirements between the parties in off-year elections (on-year elections as well, though that is not shown in the figure) strongly predicts the aggregate election outcome (with a correlation of .87). Elections with a larger proportion of GOP retirements are strongly associ- ated with a larger number of Republican Congressional losses. An excess of Democratic retirements predict a greater number of Republican Congressional wins.
The relationship is particularly consequential for 2022 because the GOP retire- ment rate usually exceeds Democratic retirements in both on- and off-years. The pre-election retirement rate for Republicans has averaged about 91⁄2 percent over the last sixty years. The retirement rate for Democrats has averaged a bit over 7 percent.
For the upcoming 2022 election, by contrast, the retirement rate for the Demo- crats is well over 10 percent, while the rate for Republicans is 6 percent.
Put differently, the GOP retirement rate advantage is over 4 percentage points in 2022 compared to a typical retirement rate disadvantage for Republicans of approximately 21⁄2 percent.
The slope in the figure predicts that 2022 will produce a disproportionate Dem- ocratic loss, because Democrat retirements are currently exceeding Republican retirements by more than 4 percentage points.
The historical pattern (the figure above) indicates that differential could produce a loss for the Democrats of nearly 40 seats. That projection is conservative be- cause it is adjusted for elections that are statistical outliers linked to unusually large presidential surges. The Democratic losses for 2022 are estimated to be nearer to 50 seats without this adjustment.
Summary
These prominent models indicate that the 118th House of Representatives will have a Republican majority. The surge and decline model, because Biden’s coattails were short in 2020 indicates the smallest (12 to 13 seats) pickup for the GOP. The president’s approval level is low and it indicates the Democrats will lose about 18 seats. The retirement model is the most challenging hurdle for the Democrats, with some 40 or more seats at risk this November.
Why the differences? The surge and decline estimation is an election structure baseline reflecting the outcome of the 2020 election. The approval measure is more dynamic and considers current political events. The retirement model, ap- plied to 2022, reflects how Democratic incumbents are estimating the political consequences of Biden’s low and declining approval level during the past year. It might be considered a behavioral indicator.
If I were to hazard a guess, and it is not much more than that, the November result will come out between a loss of 20 to 40 seats by the Democrats, with the most likely result between 20 and 30 seat losses.
This assessment has nothing to offer about the Senate races, other than to note that the relevant influences are similar and they are unlikely to generate a pre- diction or outcome that is markedly different.
