How many House votes and seats does the president's party lose in U.S. midterms? 2006–2022 official results, and what current data imply for 2026
In every U.S. midterm since 2006 the president's party lost House vote share, between 2.9 and 9.0 points of the two-party vote and between 9 and 63 seats. The size of the loss tracked the national unemployment trend closely; state-level differences in jobs, power prices and house prices explain almost nothing. Analysis, not a forecast: applied to the 2024 result and current data, the past pattern would imply about 220 to 258 Democratic seats (±10; the lowest is within error of no majority).
Questions this record answers
- How many House votes does the president's party lose in U.S. midterm elections?
- In all five midterms 2006–2022 the president's party lost two-party House vote share: 5.5 points in 2006, 9.0 in 2010, 3.4 in 2014, 5.1 in 2018 and 2.9 in 2022 (FEC official totals, 50 states). The median loss is 5.1 points.
- How many House seats does the president's party lose in U.S. midterms?
- 30 seats in 2006, 63 in 2010, 12 in 2014, 42 in 2018 and 9 in 2022, according to the Clerk of the House.
- Did unemployment affect U.S. midterm House results 2006–2022?
- Nationally the size of the loss moved with the unemployment trend (correlation −0.97 over five midterms): the largest loss, 2010, followed a rise from 5.8 to 9.6 percent; the smallest, 2022, a fall from 8.1 to 3.7 percent. Between states within the same year there was no significant relation.
- Did electricity or house prices affect U.S. midterm House results by state?
- No measurable effect: state changes in residential electricity prices and in the FHFA house price index explain almost none of the differences between states in the same midterm year (together with unemployment 1.7 percent).
- What do past midterms imply for the 2026 U.S. House election?
- Analysis, not a forecast: starting from the 2024 result (Republicans 51.3 percent of the two-party vote, 220 seats), the past five midterm losses of the president's party imply a Democratic House of about 220 to 258 seats; the unemployment pattern alone implies about 240 (each ±10 seats). Redistricting since 2024 and polls are not included.
Cite as: FACTANKER, https://factanker.com/record/midterm-house-losses-president-party-2006-2022-scenarios-2026
Analysis, not a forecast. This page relates official election results to official economic data and asks what the pattern of the last five midterms would mean if it repeated in 2026. It uses no polls, no candidate information and no district-level modelling. Every vote and seat number links to its filed record; the method and its limits are stated at the end.
1. How many House votes does the president's party lose in U.S. midterm elections (2006–2022)?
In all five midterms since 2006 the party holding the White House lost share of the two-party vote for the U.S. House compared with the election two years earlier:
| Midterm | President's party | Its two-party House vote | Change vs. two years earlier | Its House seats | Seat change |
|---|---|---|---|---|---|
| 2006 | Republican | 45.9% | −5.5 points | 202 | −30 |
| 2010 | Democratic | 46.5% | −9.0 points | 193 | −63 |
| 2014 | Democratic | 47.2% | −3.4 points | 188 | −12 |
| 2018 | Republican | 45.5% | −5.1 points | 199 | −42 |
| 2022 | Democratic | 48.7% | −2.9 points | 213 | −9 |
The median loss is 5.1 points of the two-party vote. Vote totals are the official sums of the 50 states from the Federal Election Commission; seats are the immediate result of each election as recorded by the Clerk of the House.
2. Did rising unemployment cost the president's party House votes in U.S. midterms 2006–2022?
Nationally, the two moved together. The biggest loss, 2010 (−9.0 points), came after the annual unemployment rate rose from 5.8 to 9.6 percent; the smallest, 2022 (−2.9 points), after it fell from 8.1 to 3.7 percent. Across the five midterms the correlation between the change in unemployment and the change in the president's party's vote share is −0.97: each percentage point of higher unemployment went with about 0.8 points more loss, on top of a loss of about 5.9 points when unemployment was unchanged. Five data points are few; this is an observed pattern, not a proven cause.
Between states it does not show up. Comparing states within the same midterm (239 state-midterm pairs), a state whose unemployment rose more did not swing more against the president's party (coefficient +0.58 points per percentage point, that is, if anything slightly smaller losses; not statistically significant). Unemployment rates are computed from the county records of the Bureau of Labor Statistics (LAUS), summed to states.
3. Did electricity prices affect U.S. House midterm results by state (2006–2022)?
Not measurably. States where residential electricity prices rose faster in the two years before a midterm did not swing differently from other states in the same year (+0.05 points per percent of price change, not significant). Prices are EIA residential revenue divided by sales, January to July of each year (EIA).
4. Did house prices affect U.S. House midterm results by state (2006–2022)?
No. The change in the FHFA purchase-only house price index over the two years before each midterm shows no relation to the state swing (coefficient −0.001, correlation −0.03; FHFA). Taken together, unemployment, power prices and house prices explain 1.7 percent of the differences between states within a midterm year. The economy appears in the national result, not in which states swing more.
5. What do current U.S. economic data and the 2024 result imply for the 2026 House midterms?
The starting point: in 2024 Republicans won 51.3 percent of the two-party House vote and 220 of 435 seats. In 2026 the president's party is Republican. Current data: the national unemployment rate averaged over the eleven months from September 2025 to August 2026 that BLS published (October 2025 is missing from the record) is 0.3 points above the 2024 average; residential electricity prices from January to July 2026 are on average 11.8 percent above the same months of 2024; the house price index for the second quarter of 2026 is 6.5 percent above two years earlier.
If 2026 repeated the pattern of the past five midterms:
| Scenario | Republican change | Republican two-party vote | Implied seats (D / R) |
|---|---|---|---|
| Smallest past loss (as in 2022) | −2.9 points | 48.5% | about 220 / 215 |
| Median past loss | −5.1 points | 46.3% | about 234 / 201 |
| Loss implied by the unemployment pattern (+0.3 points) | −6.1 points | 45.2% | about 240 / 195 |
| Largest past loss (as in 2010) | −9.0 points | 42.4% | about 258 / 177 |
Seats are translated from the national vote with the seats-to-votes relation of the eleven House elections 2004–2024 (about 1.4 points of seat share per point of vote share); its typical error is about ±10 seats. In every scenario based on past midterms, Democrats would win a House majority; in the mildest one only narrowly, within that error.
6. How reliable is this, and what is not in it?
- Five midterms. The national pattern rests on five observations. It describes the past; it does not predict an election.
- No polls, candidates or district maps. Several states redrew congressional districts after 2024. New maps can move seats without any change in votes; the seat estimates above do not include them.
- State vote totals include uncontested seats, where one party has no candidate. Eleven state-midterm pairs were excluded from the state comparison: four with a party missing, six with swings above 20 points and one with missing economic data.
- Interest rates are the same for every state, so they cannot be compared across states; the 10-year Treasury yield in August of the five midterm years ranged from 2.4 to 4.9 percent, with no link between its level and the size of the loss; its two-year change moved with smaller losses (correlation +0.90, five points), which mostly reflects 2010 and 2022.
Method
Vote share is each party's share of Democratic plus Republican votes in the general election for the U.S. House, 50 states (FEC "Federal Elections" 2004–2022; Clerk of the House, Statistics of the election of November 5, 2024). The swing of the president's party compares each midterm with the House election two years earlier. State comparison: least squares on the swing, with each midterm year centred separately (year fixed effects), using the change in the annual unemployment rate, the percent change in residential power prices and in the house price index over the same two years. Scenarios add each past national loss to the 2024 state results, adjust states by the (negligible) state coefficients and current data, and convert the national vote to seats; Vermont (no Republican candidate in 2024) is omitted from these rows. The unemployment row applies the national relation directly to the 2024 national share. Numbers rounded.