What Stocks, Bonds, and Cash Have Actually Delivered
Financial plans and retirement calculators run on assumed rates of return, and those assumptions should be grounded in what each asset class has actually delivered over long periods, not in optimistic guesswork. Knowing the historical range for stocks, bonds, and cash also sets realistic expectations for how bumpy the ride to that average return actually is.
The core principle: return and risk move together
Over long multi-decade periods, the major U.S. asset classes have delivered average annual returns that scale closely with how much year to year volatility each has been willing to impose on its holders. Broad stock indexes have historically delivered average annual nominal returns in roughly the 9% to 10% range, with a standard deviation near 18% to 20%, meaning individual calendar years have ranged from losses exceeding 30% to gains exceeding 30%. Long-term government bonds have historically returned roughly 5% to 6% nominal, with a standard deviation nearer 8% to 10%. Short-term Treasury bills have returned roughly 3% to 4% nominal, with very low volatility, typically well under 3%. The ordering is consistent and, once you understand the risk premium concept, unsurprising: the asset that asks investors to tolerate the widest range of single-year outcomes has also been the one that has paid, on average, the most for that tolerance.
A useful way to summarize the tradeoff in a single number is the Sharpe ratio, which measures excess return per unit of volatility: Sharpe ratio = (portfolio return − risk-free rate) / portfolio standard deviation. Using rough long-run historical figures (stocks at 9.5% return with 19% standard deviation, bills at 3.5% as the risk-free rate), stocks show a Sharpe ratio near (9.5% − 3.5%) / 19% = 6.0% / 19% = 0.32, meaning investors have earned roughly 0.32 units of excess return for every unit of volatility endured historically, a useful benchmark figure to keep in mind when a new fund's marketing materials tout an impressive-sounding Sharpe ratio of its own.
The math: two worked examples
Example 1: the compounding gap between stocks and bonds over 30 years. Starting with $10,000, a 9.5% average nominal annual stock return compounded over 30 years grows the balance by a factor of (1.095)^30, which is approximately 14.97, producing an ending balance of roughly $149,700. The same $10,000 compounded at a 5.5% average nominal bond return over 30 years grows by a factor of (1.055)^30, approximately 4.83, producing an ending balance of roughly $48,300. The stock portfolio ends with roughly three times the bond portfolio's balance, despite the annual return gap being only about 4 percentage points, a clear illustration of how a modest difference in annual compounding rate produces an enormous difference in ending wealth over a long enough horizon.
Example 2: the price of that higher return in a bad single year. Take the same stock portfolio, 9.5% average return with 19% standard deviation, and ask what a bad single year, roughly two standard deviations below the mean, would look like: 9.5% − (2 × 19%) = 9.5% − 38% = −28.5%. A $100,000 stock portfolio experiencing this scale of decline would fall to approximately $71,500 in a single year, an outcome that, while painful, sits within the range the historical volatility figure itself predicts should occur roughly once every 40 years on the downside alone (the lower half of the roughly 5% two standard deviation tail). Compare this to a bond portfolio with a 5.5% average return and 9% standard deviation experiencing its own two standard deviation bad year: 5.5% − (2 × 9%) = 5.5% − 18% = −12.5%, a materially smaller single-year decline. The higher long-run stock return in Example 1 and the deeper potential single-year decline in Example 2 are two sides of the same coin: you cannot have access to one without accepting exposure to the other.
What long-run market history shows
The pattern of stocks outpacing bonds, and bonds outpacing bills, over sufficiently long holding periods has held up with remarkable consistency across the roughly century-long span of detailed U.S. market data, and similar patterns, sometimes with different specific magnitudes, have been documented in most other major developed equity markets studied over comparably long windows. The consistency of the pattern across so many different countries, regimes, and eras is one of the more robust empirical findings in the entire discipline, even though the exact size of the premium in any given market or period has varied and remains an active area of research and debate.
What the data does not show is a smooth, predictable path to that long-run average. The historical return series for equities in particular includes multiple extended stretches, sometimes lasting a full decade or longer, where realized returns fell well short of the long-run average, interspersed with other stretches where returns ran well above it. Investors who only look at the long-run average without internalizing the volatility that produced it are systematically underestimating how difficult the path to that average has actually been to live through in real time, with real money, during the specific stretches when it was running behind.
It is worth being explicit about what these historical figures represent and what they do not. They are backward-looking averages over specific historical windows, not guarantees, and different starting and ending points within the same broad century-long dataset produce noticeably different average figures, sometimes by a percentage point or more in either direction depending on exactly which years are included. Forward-looking expected return estimates used by pension funds, endowments, and financial planners today are generally set somewhat below the very long-run historical average for equities, reflecting a view among many professional forecasters that current valuation starting points, and the shifting composition of the economy itself, make the strongest historical decades an imperfect guide to what the next few decades will deliver, even while the general ordering (stocks above bonds above bills) remains the consistent working assumption almost everyone still uses.
How this applies in real portfolios
The practical starting point for any financial plan is to use decade-plus historical averages, adjusted where appropriate for a starting valuation environment that looks unusually rich or unusually cheap relative to history, rather than either the most recent single year's return or an arbitrarily optimistic assumption pulled from nowhere. From there, the asset mix between stocks, bonds, and cash should be set primarily by how many years remain until the money is actually needed and by how much single-year volatility the investor can tolerate without abandoning the plan at the worst possible moment, since abandoning a well-designed plan during a downturn is a far larger threat to long-run outcomes than the underlying asset allocation choice itself.
For a professional managing a growing portfolio over a multi-decade career, the practical lesson from these historical figures is to lean into equities for money with a genuinely long horizon, since the compounding math in Example 1 above shows how much long-run wealth is left on the table by holding too conservative a mix out of excess caution, while reserving a bond and cash allocation sized to genuinely cushion the specific bad years illustrated in Example 2, so that a downturn does not force a sale of equities at the worst possible time to cover near-term spending needs.
Actionable breakdown
- Use decade-plus historical averages, not last year's return, for planning.
- Single years tell you almost nothing about the long-run rate.
- Adjust modestly for unusually rich or cheap starting valuations.
- Expect stock returns to swing far more than bonds or cash.
- A 20 to 30 point single-year swing is within normal range.
- Size your equity exposure so that swing is survivable.
- Match asset mix to your actual time horizon.
- Longer horizons can absorb more short-term volatility.
- Shorten the horizon assumption as a goal date approaches.
- Use Sharpe ratio to compare risk-adjusted fund performance.
- A higher return alone does not mean a better investment.
- Check the volatility that produced the return too.
It is also worth naming, explicitly, the historical volatility figures a high income professional balancing a demanding career against portfolio management should expect to actually live through, rather than only the average return, since the average is what gets quoted and the volatility is what actually gets experienced. Across the roughly century-long U.S. data window, a stock-heavy portfolio has experienced a calendar-year decline of 20% or more on multiple distinct occasions and a decline of 30% or more more than once; a bond-heavy portfolio has experienced far shallower, though not zero, single-year declines over the same span. Building a financial plan that only budgets for the average return figure, without explicitly budgeting for the emotional and practical reality of at least one and probably several of these declines occurring at some point during a multi-decade career, is building a plan on an incomplete picture.
Common pitfalls
Investors often anchor heavily to whichever decade they personally began investing in. Someone who started during an unusually strong stretch may plan around unrealistically high future returns, while someone who started right after a crash may underestimate long-run stock returns for years afterward, both distorted by a sample of one specific, personally lived period.
A second pitfall is ignoring that historical nominal return figures span periods of very different inflation environments, including stretches of double digit inflation and stretches of near-zero inflation; comparing nominal figures across those different eras without adjusting for inflation can mislead, particularly when comparing bond returns, which are more sensitive to inflation surprises than stock returns tend to be.
A third pitfall is treating the long-run average return as though it applies smoothly to any arbitrary starting and ending date an investor happens to pick, when in reality the specific window chosen, and particularly the valuation level at the start of that window, has historically had a meaningful effect on the realized return over the following one to two decades.
A fourth pitfall is comparing the historical returns of different asset classes on a mismatched basis, for instance comparing a nominal bond return against a real (inflation-adjusted) stock return, or comparing a pretax figure for one asset against an after-tax figure for another, both of which quietly bias the comparison in favor of whichever asset happened to be measured on the more flattering basis.
The single most reliable use of this historical record is not to predict the next decade's exact return, which no dataset can honestly promise, but to calibrate expectations about the shape of the ride: an asset mix chosen after genuinely internalizing both the historical average and the historical volatility behind it is far less likely to be abandoned at the worst possible moment than one chosen after looking only at the average return figure in isolation, with the volatility treated as an afterthought.
The bottom line
Stocks have historically outpaced bonds and cash by a wide, compounding margin over long periods, but only investors who can genuinely tolerate the accompanying single-year volatility have reliably captured that advantage.
Related reading: market history guide, asset allocation guide, why riskier assets have to pay you more, does time really reduce investment risk, Sharpe ratio.