TL;DR
From 1871 to 2024, US large-cap stocks delivered approximately 6.7% real annualised return including reinvested dividends. The range across rolling 30-year windows: roughly 3% to 10%.
What "155 years of returns" actually means
The longest reliable US equity series is Robert Shiller's monthly S&P composite, which begins in January 1871. It's stitched together from earlier sources for the pre-1957 period and uses the published S&P 500 thereafter. Dividends are reinvested. Inflation is measured by US CPI back to 1913 and inferred from wholesale price indices before that.
Over that 155-year window:
- Nominal CAGR: ~9.4% per year
- Real CAGR (inflation-adjusted): ~6.7% per year
- Real CAGR ex-dividends (price only): ~2.4% per year
Two-thirds of long-run real return comes from reinvested dividends. Anyone planning around price-only returns is solving the wrong problem.
The shape of the distribution
The 6.7% real long-run average hides huge variation across decades:
- Best 10-year stretches: 17%+ real (1942–52, 1949–59, 1990–2000)
- Worst 10-year stretches: −4% real (1909–19, 1929–39, 1973–82, 2000–10)
Across all rolling 30-year windows since 1871:
- 10th percentile: ~3.4% real
- Median: ~6.6% real
- 90th percentile: ~9.8% real
The 30-year window is where statistical regularity starts to assert itself. Over shorter periods, the variation is much wider.
Why 1871 is the cutoff
Two reasons:
- Data quality drops sharply pre-1871. Earlier US equity series exist (Schwert, Wilson-Jones) but they're sparse and the methodologies don't match.
- Economic regime change. US public equity markets weren't really a mass financial product until the late 19th century. Pre-1871, the modal investor was a wealthy individual buying railroad bonds, not a household holding diversified equities.
So 1871–2024 is the right window for what most people mean by "long-run stock returns".
What it doesn't tell us
The data has limits. Three notable ones:
-
Survivorship bias on a national level. The US is the strongest 155-year equity record because it didn't lose two world wars, didn't experience hyperinflation, didn't have its capital markets liquidated. Plenty of countries' equity returns from 1900 to today are far worse (Germany, Japan, Italy). A globally diversified portfolio's real return is closer to 5% than 6.7%.
-
Regime risk in the future. Past returns include the rise of the US as a global power, falling interest rates 1981–2020, globalisation, and a particular tax/regulatory regime. None of these is guaranteed to continue.
-
Valuation matters. Real returns over the next 30 years are statistically related to starting CAPE ratios. Today's CAPE is in the historical top quintile, which has historically predicted real returns closer to 3–4% for the next decade.
How FIRE planners should use it
Three honest principles:
- Plan around 4–5% real, not 6.7%. It's a more conservative assumption that hedges against the future being less generous than the past.
- Test against the worst sequences, not just the average. Our withdrawal survival tool does this for every starting point in the Shiller data.
- Diversify globally. US-only assumes the US continues to outperform. Global diversification is a small expected-return cost for a meaningful tail-risk reduction.
The data is the foundation of every honest retirement projection. It's also a reminder that long-run averages hide long stretches of pain.
Frequently asked questions
- Where does the Shiller data come from?
- Robert Shiller publishes the dataset free on his Yale homepage. It combines published S&P 500 data since 1957, S&P Composite earlier, and Cowles Commission data for the pre-1926 period.
- Why not use older non-US data?
- Quality and comparability. The Dimson-Marsh-Staunton dataset extends to other countries since 1900, and we use it where global comparisons matter. But pre-1900 non-US data is sparse and inconsistent.
- Does 6.7% real include all costs?
- No — that's gross of fund fees, trading costs, and taxes. A modern index investor pays 0.1–0.5% of that back in costs and taxes; an investor with 1%+ fees gives back materially more.
Stress-test your own FIRE plan
FIRE Wealth OS runs your savings rate and expenses against every historical market starting point since 1871. Free to use, no card required.