What the Data Says About Average Investment Returns Over 30 Years

Updated on April 29, 2026

Every retirement calculator asks you the same uncomfortable question: what annual return do you expect? Type in 10% and your projections look rosy. Drop to 5% and suddenly you need to work until 73. The number you enter matters enormously — yet most people guess. They pick something that sounds reasonable, or they use the default the calculator supplies without questioning it. This is worth examining more carefully, because there is actually quite a lot of historical data on this question, and some of it contradicts the assumptions people carry around in their heads.

Let us go through what the evidence actually shows, asset class by asset class, and then talk about how to translate historical data into projections that are honest rather than wishful.

U.S. Equities: The Number Everyone Quotes Is Technically Correct and Practically Misleading

The S&P 500's long-run average annual return is frequently cited as around 10% in nominal terms. The Ibbotson SBBI data going back to 1926 supports roughly that figure for large-cap U.S. stocks. But that headline number conceals three important complications.

First, inflation erodes it substantially. In real (inflation-adjusted) terms, U.S. large-cap equities have returned closer to 6.5–7% per year over the same period. The difference between 10% nominal and 6.7% real is not semantic — over 30 years, a $100,000 investment grows to roughly $1.74 million at 10% nominal but has purchasing power closer to $700,000 in today's dollars at 6.7% real. Which number matters to you depends on whether your retirement expenses will be fixed in nominal dollars or in real ones. (They will be real ones.)

Second, sequence matters enormously. The arithmetic average return and the compound (geometric) annual return are not the same thing when returns fluctuate. A year of +30% followed by a year of -23% gives you an arithmetic average of +3.5% but a compound return of 0%. For the S&P 500, the compound (CAGR) figure over most 30-year windows runs about 0.5–1.5 percentage points lower than the arithmetic average. Vanguard's research has repeatedly highlighted this gap, noting that projection calculators using arithmetic averages systematically overstate expected ending wealth.

Third, your 30 years might not look like the average 30 years. Rolling 30-year periods for U.S. equities from 1926 to 2020 show real CAGRs ranging from roughly 4.3% (the cohort that started near the 1929 peak) to around 10.8% (those who started in the early 1970s and rode the 1980s–90s bull market). The median is close to 7% real, but the range is wide enough that "use the average" is not actually a low-risk assumption.

International Equities: The Diversification Discount That Is Actually a Feature

The Dimson, Marsh, and Staunton Global Investment Returns Yearbook — the most comprehensive long-run international dataset available — covers 35 countries from 1900 onward. What it shows is humbling for U.S.-centric investors. The U.S. had one of the best equity market track records of any country in the 20th century. This is not a coincidence you can extrapolate forward; it partly reflects survivorship bias (markets that did badly, like Russia in 1917 or China in 1949, simply ceased to exist for foreign investors).

When researchers look at a market-cap-weighted world equity index over the past century, the real return comes in around 5–5.5% per year — meaningfully lower than the U.S.-only number. For a 30-year projection using a globally diversified equity portfolio, a real CAGR assumption in the 5–7% range reflects the evidence better than 10%.

Bonds: A Tale of Two Eras

Government bonds are where the recent historical record is most treacherous to extrapolate from. From 1982 to 2021, a near-40-year bull market in bonds produced total returns that vastly exceeded what the coupon yield alone would have predicted, because falling interest rates push bond prices up. The Bloomberg U.S. Aggregate Bond Index delivered roughly 6–7% annualized over this period. That is not a number to put in a forward-looking model.

Over genuinely long periods — the full Ibbotson dataset — intermediate U.S. Treasuries have returned about 3.3% nominally, or close to 0.4% in real terms. Investment-grade corporate bonds have done somewhat better, approximately 1–2% real over the very long run. Post-2022, with yields in the 4–5% range on Treasuries, the near-term nominal return on investment-grade bonds looks better than the trailing decade, but real returns still depend on where inflation settles.

For projection purposes: a real return of 1–2% on a diversified bond portfolio is defensible. Anything much higher requires an explanation grounded in current yield levels rather than recent history.

The Balanced Portfolio: What 60/40 Actually Delivered

The classic 60% equity / 40% bond allocation is a useful benchmark because it reflects what a large fraction of retirement-oriented investors actually hold in some form. Over the 30-year period ending December 2023, a simple U.S.-based 60/40 portfolio delivered roughly 8.5% nominal CAGR, or about 5.5–6% real. This is a period that included two severe equity bear markets (2000–2002 and 2008–2009) as well as the 2022 bond-and-stock drawdown, so it is not a cherry-picked optimistic run.

For someone using a retirement calculator today, a 5–6% real return assumption for a balanced portfolio is within the range that historical data supports. A 7% real assumption is at the high end. An 8%+ real assumption is unsupported by the broad international evidence and should require a strong explicit justification.

What This Means When You Sit Down With a Calculator

Here is what most financial planning software does not make obvious: the return assumption you enter should match the return type the calculator uses. If it compounds your contributions using a single annual figure, that figure should be a CAGR (compound annual growth rate), not an arithmetic average. If the calculator asks for a "nominal" return, you need to also tell it your inflation assumption, or the output will overstate your real purchasing power. These distinctions trip up people who are otherwise fairly financially literate.

A few practical calibration points:

  • For an all-equity, U.S.-heavy portfolio: 6–7% real CAGR is historically grounded. Using 8% real is optimistic; 10% real has essentially no empirical support.
  • For a globally diversified equity portfolio: 5–6.5% real is better supported.
  • For a 60/40 balanced portfolio: 4.5–6% real has strong backing across most historical periods.
  • For a conservative (heavy bond) portfolio: 2–4% real is realistic given current yield levels.

One technique that gets too little use: run your projections with three scenarios — an optimistic case (75th percentile historical outcome), a base case (median), and a conservative case (25th percentile). The difference in how much you need to save or how long your money lasts is often larger than people expect. The exercise forces honesty about the range of outcomes rather than collapsing everything into a single line on a graph.

The Fees and Taxes That Quietly Eat the Return

Historical return data is gross of fees and taxes unless specifically stated otherwise. If your portfolio incurs 0.8% per year in fund expenses, advisory fees, and transaction costs — and that is conservative for many retail investors using actively managed funds — you need to subtract that from your projected return. Over 30 years, the compounding effect of an extra 0.8% drag is not trivial. On a $500,000 portfolio growing at 6% versus 5.2%, the 30-year difference exceeds $350,000.

Tax drag is harder to generalize because it depends on account type (taxable vs. IRA vs. 401(k)) and income tax rates. But in a taxable account, a rough rule of thumb from research by Vanguard and others is that taxes subtract another 0.5–1.5% per year from effective returns, depending on turnover and tax bracket. A pre-tax return assumption of 6% real in a taxable account might translate to 4–5% real after taxes.

Setting Your Own Assumption

The honest answer is that no one knows what the next 30 years will return. Valuations matter for long-horizon forecasts — research by John Campbell, Robert Shiller, and others shows that the cyclically adjusted price-to-earnings ratio (CAPE) has modest but meaningful predictive power over 10- and 15-year horizons. As of mid-2020s, elevated U.S. equity valuations have led many forecasters (including research teams at Vanguard, GMO, and Research Affiliates) to project U.S. equity real returns in the 2–5% range for the coming decade, not the historical 6–7%.

This does not mean those forecasts are right. It means that anchoring purely on the rear-view mirror is its own kind of mistake. The sensible approach is to use the historical median as a central estimate, build in a meaningful downside scenario that is 1.5–2 percentage points lower, and make sure your savings plan is robust enough to survive the downside without requiring the upside. A calculator that can only tell you whether you will be okay if history repeats exactly is not giving you a complete picture. The data is a starting point, not a guarantee.