Great Expectations, Meet Arithmetic

Global Perspectives | August 2026

• American individual investors say they expect long-run stock returns of nearly 13% above inflation. The median forecast of major investment firms is less than 4.0%.
• A simple valuation identity used by Nobel laureate Peter Diamond in 1999 — and vindicated over the following decade — explains why today’s prices favor the lower figure.
• An AI-transformed economy could change the arithmetic, but only with sustained growth acceleration beyond anything in the modern historical record. Diversification remains the prudent response.

WHAT RETURNS SHOULD YOU EXPECT FROM STOCKS?

Ask what the stock market will return over the long run, and the answer depends entirely on whom you ask. In the most recent Natixis global survey, American individual investors — people with at least $100,000 already invested — said they expect long-term returns of 12.6% per year above inflation. The financial advisors who serve them call 7.1% realistic, a figure close to the actual historical record: US stocks have delivered about 6.9% per year, after inflation, since 1928. A live bipartisan Senate proposal to shore up Social Security by investing borrowed funds in equities assumes 6.5%.¹ The Social Security actuaries who evaluate such ideas run official scenarios at 5.8% and 4.8%.² And the major investment houses that publish long-term capital market assumptions cluster, at the median, near 3.7% (Figure 1).

The wide range of return expectations can be visualized on a ladder, ranging from retail exuberance at the top to institutional sobriety at the bottom (Figure 2). The rungs cannot all be right. The gap between the top and bottom, compounded over a 30-year retirement plan, is not a rounding error; it is the difference between a nest egg growing roughly thirty-five-fold and merely threefold. So, it is worth asking which part of the ladder the evidence supports — and it turns out a Nobel laureate answered a version of this question, under strikingly similar circumstances, a generation ago.

AN OWNER’S ARITHMETIC

In September 1999, the MIT economist Peter Diamond examined whether the 7% assumption then used in Washington’s Social Security debate could be squared with the stock market’s lofty prices.³ His tool was disarmingly simple. Strip away the daily noise and owning the stock market resembles owning a rental property or a family farm: the long-run return is the cash it pays you now, plus the growth in those payments over time. This is the logic of the classic model of stock valuation known as the Gordon growth model, first formalized in 1956.⁴ For the market as a whole, the cash is dividends plus net share buybacks — today about 2.5% of the total US equity market’s value — and the growth of corporate cash flows cannot, over long horizons, outrun the economy that generates them (Figure 3).⁵ Profits cannot compound at 6% forever in an economy growing at 2%; eventually they would swallow the entire national income.

This is less a theory than an accounting constraint, and it pins the answer down. A cash yield of roughly 2.5% plus the 1.5% – 2% long-run real growth projected by government and private forecasters alike implies a sustainable return of roughly 4% – 4.5%. The Center for Retirement Research at Boston College, running the same logic this spring with slightly more conservative inputs, arrived at 3.5% – 4%.⁶ Neither figure is anywhere near 7%. Notice where they land on the ladder: right among the investment houses’ published numbers. The experts are not being dour; they are doing the arithmetic.

DIAMOND THEN, WASHINGTON NOW

What makes Diamond’s 1999 exercise worth retelling is what happened next. He calculated that, given the prices of that day, the market would need to fall roughly 35% – 45% in real terms over the following decade for 7% returns to be sustainable thereafter. Over the subsequent 10 years — the stretch investors remember as the lost decade — the market’s real price decline landed inside his range almost exactly. He made no claim about timing or trigger. He simply showed that the assumptions in circulation were internally inconsistent and insisted that one of them would have to give. One did.

The circumstances that prompted his analysis have now largely reassembled. The Shiller price-earnings ratio recently crossed 40 for only the second time in a century — the first was the dot-com peak (Figure 4). The dividend yield has slipped to 1.1%, an all-time low. And Washington is once again debating equity returns with real money attached: the Cassidy-Kaine proposal would have the federal government borrow $1.5 trillion to invest in stocks, with a cumulative $26.6 trillion in borrowing riding on its assumed 6.5% real return.

We updated Diamond’s table with 2026 inputs, anchoring on that 6.5% figure — which sits conveniently close to both the historical record and advisors’ expectations. The result: for today’s market to deliver 6.5% real returns on a sustainable basis, prices would first need to decline about 38% in real terms over the coming decade (Figure 5). For the full historical 6.9%, about 43%. For the actuaries’ 5.8% scenario, about 28%; for their 4.8%, roughly 6% — close to break even. And at 4%, no decline at all is required: today’s prices are entirely consistent with stocks earning about 4% above inflation from here, indefinitely.
There is an equivalent way to see the same math, courtesy of the Boston College researchers: apply a steady 6.5% return to today’s prices with no correction, and the stock market’s value grows from about twice US GDP now to more than 13 times GDP by 2100. Nobody believes that (Figure 6).

The table’s message is not that a crash is coming. It is that the market does not need to crash to disappoint. It merely needs to deliver what its own price implies — the quiet, cumulative shortfall of the ladder’s bottom rungs — unless something changes the growth side of the equation. Which brings us to the argument of the moment.

THE THIRD DOOR

In 1999, Diamond noted that faster economic growth could resolve the inconsistency, then set the idea aside; nobody was forecasting it. In 2026, that unexplored possibility is the entire bull case. If artificial intelligence genuinely transforms productivity, the growth number changes and the arithmetic cooperates. So rather than dismiss it, invert the table and ask: how fast would the economy need to grow, permanently, for today’s prices to deliver each return with no decline at all?

The answer is the return minus the cash yield (Figure 7). A 6.5% return requires real growth of 4% a year, forever — against a postwar average of about 3%, a recent trend near 2%, and official long-run projections around 1.5%. And even that understates the miracle required, because history shows the earnings of today’s companies lag the growth of the whole economy by roughly two percentage points a year, as new firms — many not yet founded — capture their share of the future.⁷ Much of the internet’s eventual value, recall, accrued to companies that did not exist in 1999.

None of this requires believing AI will fail. Vanguard’s current outlook makes the point precisely: it assigns a 60% probability to the US economy achieving 3% real growth in the coming years — genuine AI optimism — and still projects only 4% – 5% nominal returns for US equities, on the grounds that expectations already embedded in prices are high and that creative destruction erodes incumbents’ profits.⁸ A booming economy and a disappointing index can coexist. For investors in railroads, electricity, and aviation — transformative technologies, all — they frequently did.

CHOOSING A RUNG

Where does this leave an investor staring at the ladder? The top rung — the 12.6% of popular expectation — has no support in either history or arithmetic; its nearest precedent is a 1997 survey, cited in Diamond’s own footnotes, in which mutual fund investors expected 34% a year for the coming decade. The middle rungs — history’s 6.9%, the Senate’s 6.5% — require either a price decline on the scale of the lost decade or extraordinary growth. The bottom rungs — the experts’ roughly 3.5% – 4.5% — require nothing but today’s prices and ordinary growth. Prudent planning starts there and treats anything better as a pleasant surprise.

None of this is a market call, and the record demands humility about timing: valuations looked stretched for years before 2000, and investors who fled early paid dearly. What the arithmetic does support is scrutiny of any plan — personal or national — that quietly assumes the historical average from today’s starting point. It also points somewhere constructive. The same method for projecting 3% from US large caps projects meaningfully more from US small caps, value stocks, and international markets, where starting yields are higher and less perfection is priced in.

Not surprisingly, realistic equity return assumptions are more supportive of bond allocations as well. When one asset’s math requires extraordinary growth, the time-honored response is not to abandon it, but to own other things too. Great expectations are a fine literary theme. Arithmetic is a better investment plan.

 

William P. Sterling, Ph.D.,

Global Strategist

 

1 The bipartisan proposal by Senators Bill Cassidy (R-LA) and Tim Kaine (D-VA) is analyzed in Chen, Munnell & Aubry, “Can Equity Investments Help Social Security’s Finances?”, CRR at Boston College, IB 26-10, May 2026.

2 Social Security Administration, Office of the Chief Actuary. “Provisions Affecting Trust Fund Investment in Equities.” Social Security Administration, www.ssa.gov/oact/solvency/provisions/investequities_summary.html. Accessed July 23, 2026.

3 Peter Diamond, “What Stock Market Returns to Expect for the Future?”, CRR IB No. 2, September 1999.

4 Gordon, M. J., & Shapiro, E. (1956). Capital equipment analysis: The required rate of profit. Management Science, 3(1), 102–110.

5 Estimates vary with measurement choices. Federal Reserve financial accounts data for the nonfinancial corporate sector imply roughly 2.3%; normalizing net buybacks with ten-year rolling averages relative to market capitalization yields about 2.6%; and S&P Dow Jones Indices data on dividends and repurchases, net of share issuance, suggest 2.5 to 2.7%. We use 2.5% throughout.

6 Chen, Munnell & Aubry, “Can Equity Investments Help Social Security’s Finances?”, CRR at Boston College, IB 26-10, May 2026.

7 Bernstein, W. J., & Arnott, R. D. (2003). Earnings growth: The two percent dilution. Financial Analysts Journal, 59(5), 47–55.

8 Vanguard. 2025. AI Exuberance: Economic Upside, Stock Market Downside. Vanguard Economic and Market Outlook for 2026. Valley Forge, PA: Vanguard, December 10.