Ray Fair, Yale · Economy line
Fair gives Republicans 51.0% of the 2028 popular vote
That is a 61% chance of winning it. In the backtest, this model’s typical miss was ±3.4 points, so 80% of outcomes would land between 46.4% and 55.6%.
Two-party popular vote, not the Electoral College. If the economy follows the Fed’s projections: FOMC median projections fill the quarters through mid-2028; approval is today’s. Source: FRED/ALFRED (BEA, BLS, Census), FOMC projections; model fits by this site. As of October 17, 2026.
Show the numbers
| Date | Share | 80% range | Chance |
|---|---|---|---|
| October 14, 2026 | 51.0% | 46.4%–55.6% | 61.3% |
| October 15, 2026 | 51.0% | 46.4%–55.6% | 61.3% |
| October 16, 2026 | 51.0% | 46.4%–55.6% | 61.3% |
| October 17, 2026 | 51.0% | 46.4%–55.6% | 61.3% |
What pushes it up or down
Two-party popular vote, not the Electoral College. Source: FRED/ALFRED (BEA, BLS, Census), FOMC projections; model fits by this site. As of October 17, 2026.
Show the numbers
| Input | Points |
|---|---|
| Typical party-in-power edge | +1.6 |
| No incumbent on the ballot | −2.0 |
| Time the party has held power | +1.9 |
| Quarters of strong growth | −1.3 |
| Which party holds the White House | +0.6 |
| Growth per person (latest 3 quarters) | +0.1 |
| Inflation over the term | +0.1 |
Today’s inputs against 1948–2024
If the economy follows the Fed’s projections: FOMC median projections fill the quarters through mid-2028; approval is today’s. Source: FRED/ALFRED (BEA, BLS, Census), FOMC projections; model fits by this site. As of October 17, 2026.
Show the numbers
| Year | Growth per person | Inflation over the term | Strong-growth quarters |
|---|---|---|---|
| Today | 1.9 | 2.9 | 2.0 |
| 1948 | 2.2 | — | — |
| 1952 | 0.5 | 2.4 | 7.0 |
| 1956 | −1.5 | 1.9 | 6.0 |
| 1960 | 3.7 | 2.2 | 5.0 |
| 1964 | 3.0 | 1.3 | 8.0 |
| 1968 | 4.9 | 3.0 | 9.0 |
| 1972 | 6.2 | 4.4 | 5.0 |
| 1976 | 4.9 | 7.8 | 5.0 |
| 1980 | −3.7 | 8.2 | 5.0 |
| 1984 | 5.7 | 5.3 | 6.0 |
| 1988 | 2.0 | 3.1 | 6.0 |
| 1992 | 1.0 | 3.8 | 0.0 |
| 1996 | 1.8 | 2.3 | 3.0 |
| 2000 | 3.3 | 1.7 | 8.0 |
| 2004 | 2.9 | 2.0 | 2.0 |
| 2008 | 0.2 | 2.9 | 3.0 |
| 2012 | 1.0 | 1.6 | 1.0 |
| 2016 | 1.0 | 1.4 | 2.0 |
| 2020 | −4.9 | 1.8 | 2.0 |
| 2024 | 1.6 | 4.5 | 4.0 |
How it works
Ray Fair has published this equation since 1978. Voters reward the party in power for growth in the election year and punish it for inflation over the term.
It also counts the quarters of strong growth (over 3.2% a year per person), whether a sitting president is on the ballot, and how many terms the party has held the White House.
Here it is re-fitted on 1948–2024 with the data as it was published before each election. Fair’s war-year term is dropped because only 1948 uses it.
If the economy follows the Fed's projections (FOMC median, 2026-09-16): quarters after 2026 Q2 grow at the projected rate for their year; PCE inflation stands in for the GDP deflator; income per person grows with GDP per person.
The fitted equation
| Term | Coefficient |
|---|---|
| Constant | 50.590 |
| Democratic incumbent party | −1.239 |
| Growth per person, election year | 0.958 |
| Inflation over the term | −0.852 |
| Strong-growth quarters | 0.547 |
| President running | 3.720 |
| Terms the party has held power | −3.264 |
Its record
Predicting each election with a version fitted without it, the model’s typical error is ±3.4 points and it called the popular-vote winner in 14 of 20 elections. It missed 1948, 1960, 1968, 1988, 2016, 2020. See every election.