Do markets love gridlock?
Myth. Since 1927, divided years averaged +6.8% above cash, against +10.9% for one-party years. In the last 20 years: a near tie, +11.1% against +10.9%.
Stocks do better in years when power in Washington is split than in years when one party holds the White House, the House and the Senate.It went the other way: divided years trailed one-party years by 4.1 points on average, though the likely range includes zero. In the last 20 years it was a near tie (+0.3 points). Gridlock came out behind in both halves of the record and did better in only 5 of 47 industries.
What do people believe about gridlock?
The belief is that stocks do better when power in Washington is split between the two parties.
Gridlock is when the White House and Congress can't agree, so few big laws get passed. The idea is that markets like the quiet.
The line “markets love gridlock” comes back at every election. It seems to have caught on in the 1990s, during a boom with power split.
What exactly did we test?
We compared years when power was split with years when one party ran everything.
A year counts as divided if, on the first of July, the president's party lacked a majority in the House, the Senate or both. The labels come from the House and Senate historians' official tables.
From 1927 to 2025, that gives 49 divided years and 50 one-party years. In a one-party year, the same party held the White House, the House and the Senate.
We used the whole US stock market, with data through August 2026. Every return is measured above what cash in Treasury bills earned that year. Gaps between averages are in percentage points, or points for short.
What did we find?
Divided years were not the better ones: they trailed one-party years since 1927, and ran a near tie in the last 20 years.
Chart data
| Period | Divided government | One party in charge |
|---|---|---|
| All years | +6.8% | +10.9% |
| 1927–1988 | +4.0% | +11.1% |
| 1989–2025 | +9.7% | +10.4% |
| Before 1995 | +4.9% | +10.6% |
| Since 1995 | +9.6% | +11.9% |
| Last 20 years | +11.1% | +10.9% |
Divided years averaged +6.8% above cash, against +10.9% for one-party years: a gap of 4.1 points the other way from the saying. They also beat cash a little less often: 69% of them did, against 72% of one-party years.
But don't flip the saying around. The likely range of the gap runs from -12.1 to +3.8 points, which includes zero. With about a hundred years and few changes of power, a gap this size could easily be luck.
In the last 20 years, the gap was +0.3 points, a near tie that tilts slightly toward gridlock. Divided years averaged +11.1% above cash, against +10.9%. Of the 12 divided years, 92% beat cash, against 75% of one-party years.
How did each divided year do?
About half: 25 of 49 divided years beat the average one-party year, and 8 of 12 in the last 20 years (one-party: 5 of 8).
Chart data
| Year | Value | Average one-party year | Lit |
|---|---|---|---|
| 1932 | -9.7% | +10.9% (up to 2005) | no |
| 1947 | +3.0% | +10.9% (up to 2005) | no |
| 1948 | +1.1% | +10.9% (up to 2005) | no |
| 1955 | +23.8% | +10.9% (up to 2005) | yes |
| 1956 | +5.8% | +10.9% (up to 2005) | no |
| 1957 | -13.1% | +10.9% (up to 2005) | no |
| 1958 | +43.5% | +10.9% (up to 2005) | yes |
| 1959 | +9.7% | +10.9% (up to 2005) | no |
| 1960 | -1.4% | +10.9% (up to 2005) | no |
| 1969 | -17.5% | +10.9% (up to 2005) | no |
| 1970 | -6.5% | +10.9% (up to 2005) | no |
| 1971 | +11.8% | +10.9% (up to 2005) | yes |
| 1972 | +13.1% | +10.9% (up to 2005) | yes |
| 1973 | -26.1% | +10.9% (up to 2005) | no |
| 1974 | -35.9% | +10.9% (up to 2005) | no |
| 1975 | +32.6% | +10.9% (up to 2005) | yes |
| 1976 | +21.9% | +10.9% (up to 2005) | yes |
| 1981 | -18.0% | +10.9% (up to 2005) | no |
| 1982 | +11.1% | +10.9% (up to 2005) | yes |
| 1983 | +13.8% | +10.9% (up to 2005) | yes |
| 1984 | -6.1% | +10.9% (up to 2005) | no |
| 1985 | +24.9% | +10.9% (up to 2005) | yes |
| 1986 | +10.3% | +10.9% (up to 2005) | no |
| 1987 | -3.9% | +10.9% (up to 2005) | no |
| 1988 | +11.7% | +10.9% (up to 2005) | yes |
| 1989 | +20.5% | +10.9% (up to 2005) | yes |
| 1990 | -14.0% | +10.9% (up to 2005) | no |
| 1991 | +29.3% | +10.9% (up to 2005) | yes |
| 1992 | +6.2% | +10.9% (up to 2005) | no |
| 1995 | +31.3% | +10.9% (up to 2005) | yes |
| 1996 | +16.0% | +10.9% (up to 2005) | yes |
| 1997 | +26.0% | +10.9% (up to 2005) | yes |
| 1998 | +19.6% | +10.9% (up to 2005) | yes |
| 1999 | +20.7% | +10.9% (up to 2005) | yes |
| 2000 | -17.5% | +10.9% (up to 2005) | no |
| 2001 | -15.0% | +10.9% (up to 2005) | no |
| 2002 | -22.7% | +10.9% (up to 2005) | no |
| 2007 | +1.1% | +10.9% (last 20 years) | no |
| 2008 | -38.2% | +10.9% (last 20 years) | no |
| 2011 | +0.6% | +10.9% (last 20 years) | no |
| 2012 | +16.3% | +10.9% (last 20 years) | yes |
| 2013 | +35.0% | +10.9% (last 20 years) | yes |
| 2014 | +11.7% | +10.9% (last 20 years) | yes |
| 2015 | +0.2% | +10.9% (last 20 years) | no |
| 2016 | +13.4% | +10.9% (last 20 years) | yes |
| 2019 | +28.4% | +10.9% (last 20 years) | yes |
| 2020 | +23.6% | +10.9% (last 20 years) | yes |
| 2023 | +21.7% | +10.9% (last 20 years) | yes |
| 2024 | +19.7% | +10.9% (last 20 years) | yes |
Each column is one divided year. The dashed line is the average one-party year of its span: +10.9% from 1927 to 2005, and +10.9% in the last 20 years. A column above the line is a divided year that beat it.
Up to 2005, 17 of 37 divided years beat it, fewer than half. One-party years beat their own average 22 times in 42, and 27 times in 50 over the whole record. If gridlock lifted stocks, divided years should have cleared the line more often than one-party years, not less.
The recent 8 of 12 is close to the one-party rate, and a dozen years is too few to call that a lean either way.
Divided years had big swings too. The best divided year was 1958, at +43.5% above cash. The worst divided year was 2008, when stocks ended at -38.2% against cash. That one year pulled the recent divided average back to a near tie.
What happened when the saying caught on?
The saying seems to have caught on in eight divided years, 1995 to 2002, which began with a boom and ended in a bust.
From 1995 to 1999, a Democratic president shared power with a Republican Congress. Those 5 years averaged +22.7% above cash.
Then the dot-com bubble burst. The next 3 years, 2000 to 2002, all still counted as divided, averaged -18.4%.
Over all 8 years, the run averaged +7.3% above cash, less than the average one-party year (+10.9%). The boom was real; the full run didn't back the saying. One eight-year stretch can't settle it either way. In the last 20 years, divided years averaged +11.1%, against +10.9% for one-party years.
Does it hold in every decade and industry?
No: divided years came out ahead in only 3 of 10 decades, and the last 20 years were close to a tie (+0.3 points).
Chart data
| Period | Divided years' edge |
|---|---|
| 1930s | -16.0 |
| 1940s | -10.3 |
| 1950s | -7.7 |
| 1960s | -18.6 |
| 1970s | -0.4 |
| 1980s | -14.9 |
| 1990s | +14.8 |
| 2000s | -35.2 |
| 2010s | +4.4 |
| 2020s | +16.4 |
| Last 20 years | +0.3 |
Divided years led in the 1990s, 2010s and 2020s so far, and trailed in the other decades, the 1970s only just. Some decades held only one or two divided or one-party years, so a single year could decide them.
In both halves of the record, gridlock trailed. From 1927 to 1988, divided years trailed one-party years by 7.1 points. From 1989 on, they trailed by 0.7 points, close to a tie.
We also checked 47 US industries with enough history. Divided years did better in only 5 of them. Industries move together, so these aren't independent votes.
Did the president's party matter more?
In the history it looks that way, but the groups are small, so this is a pattern in the record, not a tested finding.
Chart data
| Period | All years since 1927 | Last 20 years |
|---|---|---|
| Dem. president, one party | +12.3% | +12.1% |
| Dem. president, divided | +13.7% | +14.8% |
| Rep. president, one party | +7.2% | +9.6% |
| Rep. president, divided | +3.5% | +3.7% |
Years with a Democratic president averaged +12.3% above cash when the party also held Congress, over 36 years. When it didn't, they averaged +13.7%, over 16 years.
Years with a Republican president averaged +7.2% with Congress on side, over 14 years. Without it, they averaged +3.5%, over 33 years.
Most divided years (33 of 49) had a Republican president, so that group weighs most in the divided average.
The last 20 years looked similar. Divided years averaged +14.8% under Democratic presidents and +3.7% under Republican ones; one-party years +12.1% and +9.6%. Each group held 4 to 8 years.
Please don't read politics into this: the groups are small and don't show that either party is better for stocks.
Does gridlock matter in the month after an election?
Not clearly: divided results averaged +1.26% above cash in that month, one-party results +0.75%, and the last 20 years went the other way.
The window is the 21 trading days after the last close on or before each federal election, 1926 to 2024. Divided means the president's party didn't hold both chambers once the new Congress and president took office.
Chart data
| Period | Divided result | One-party result |
|---|---|---|
| Month after the election all years | +1.26% | +0.75% |
| Month after the election last 20 years | +0.92% | +1.72% |
That month usually beat cash: 65% of 23 divided results and 63% of 27 one-party ones.
Gridlock came out ahead by 0.50 points, but the likely range, -2.45 to +3.49 points, includes zero. A single month swings a lot, and there are few elections.
In the last 20 years it went the other way, +0.92% against +1.72%, but from just 6 and 4 elections. Its likely range, -8.89 to +11.02 points, is wide, so that could easily be luck too.
Why might it seem to work?
Mostly because the story makes sense.
Timing helps too. The saying seems to have caught on in the late-1990s boom, and a boom sticks in the memory longer than the bust after it.
Markets, it turns out, are not that sentimental about who runs the House.
What's the story in numbers?
Since 1927 there were 49 divided years and 50 one-party years. Divided years averaged +6.8% above cash, one-party years +10.9%; in the last 20 years, +11.1% against +10.9%. A divided year beat the average one-party year 25 times in 49, and 8 in the last 12. One-party years beat their own average 27 of 50 times, and 5 of the last 8. Divided years led in 3 of 10 decades.
Three things to take away. Stocks did well under both kinds of government: both averaged well above cash. Whether power was split didn't tell you which years would be good. And one vivid stretch isn't a pattern: the boom when the saying seems to have caught on ended in a bust.
What does this mean for you?
This is history, not a forecast: who controls Washington hasn't reliably told you which years would be good for stocks.
A myth doesn't mean the opposite is true. One-party years did better on average, but not reliably, so betting on them would be its own untested idea.
Trading on gridlock would also have grown less. Holding stocks only in divided years, with cash otherwise, grew +5.5% a year in total return, against +9.9% for staying invested. In the last 20 years of data: +7.0% against +11.5%.
Its worst fall was smaller, -65.0% against -83.9%, for the same reason it grew slowly: it was in stocks only 47% of the time. In the last 20 years, its worst fall was -50.9%, against -54.7%.
One catch: in 2001 the Senate changed hands in June, so that year's label wasn't known in January. Every other label was known from the election before.
What's in the small print?
Data: the US total stock market and one-month Treasury bills from Kenneth R. French's data library (Tuck School of Business at Dartmouth), daily from July 1926 to August 2026, and 49 US industry portfolios from the same library. Party control comes from the House and Senate historians' official tables.
Method: each year runs from the last close of the year before to the last close of the year. Its label is the one in force on the first of July, mid-year. A Senate tie goes to the vice-president's party, and the 2001 Senate switch is included. 1926 is left out because our data start in July that year.
Returns are compounded, before taxes and, except for the timing example, above T-bills. The timing example pays a twentieth of a percent per switch. The likely ranges are 95% bootstrap intervals.
Divided and one-party years come in runs, so they aren't independent draws. That's one more reason the likely range is wide.
Test many beliefs and about 1 in 20 with nothing behind it will pass the usual 95% standard by luck alone. So the main ranges are also shown at 99.67%, which allows for testing 15 calendar beliefs.
The gap's likely range is -12.1 to +3.8 points at 95%, or -16.1 to +8.0 points at 99.67%. Since 1995, when the saying seems to have caught on, divided years trailed by 2.4 points, with a likely range of -14.6 to +9.8 points.
The year-by-year, decade and 1995–2002 figures describe the same data; they aren't part of the verdict. Both kinds of year are compared with the average one-party year of their span: 1927 to 2005, or the last 20 years, 2006 to 2025. The decade chart leaves out the 1920s, which had no divided year, and its 2020s run to 2025.
Other checks: measured without subtracting cash, the gap was -2.0 points. For the average industry, with each one weighted equally, it was -8.3 points. The breakdown by the president's party isn't part of the verdict.
Month-after check: in midterm years, its windows match our midterm page.
Its labels show who took office; a few races were settled after the window began.
What else did we test?
See all the calendar effects we tested, or how we test a myth.