The Stock Market Rewards Optimists, But Not the Naïve Ones

Elroy Dimson co-wrote the book that financial advisors quote when they want a client to keep buying stocks. Triumph of the Optimists documented 101 years of global market history and showed that, across most countries and most long horizons, holders of productive risk assets were rewarded for their patience. The title became shorthand for “stocks always win in the long run.”

That is not what Dimson actually argues. On his May 2026 Rational Reminder appearance with Ben Felix and Cameron Passmore, Dimson spends most of the conversation warning investors away from the cheap version of his own thesis. He says U.S. history is an outlier, that easy-to-access data overstates how good things were, that economic growth does not flow cleanly to public shareholders, and that the equity risk premium most people quote is too high. His optimism has conditions attached.

Dimson’s conditions have direct consequences for DIY financial plans. The stock market does reward optimists. It does not reward naïve ones.

Financial History Is the Input to Every Plan

Dimson, Paul Marsh, and Mike Staunton built the Dimson-Marsh-Staunton (DMS) global returns database starting in the late 1990s. The 2026 UBS Global Investment Returns Yearbook now covers annual returns on equities, bonds, bills, inflation, currencies, and gold across 35 markets from 1900 through 2025 (UBS Yearbook 2026 public summary). That coverage matters because a financial plan inherits the assumptions baked into whatever historical record it pulls from. A retirement projection built on U.S. 1900-2025 returns will land in a different place than one built on the global DMS record.

For the dataset itself in detail: Do 200 Years of Stock Returns Still Matter? unpacks survivorship corrections in the pre-1862 U.S. record, and The Real Return of Stocks derives the 5 percent global real planning anchor. The work that follows here is about the interview itself: what Dimson actually argues and how DIY investors should use it.

Easy Data Bias: The Missing Mess Is Where the Risk Lives

Dimson uses the term easy data bias on the podcast: the easiest data to collect tends to start after the most chaotic period, exclude markets that failed, and assume currencies that survived. He gives the example of U.K. and German equity series that historically began after wartime turmoil. A backtest that starts in 1948 understates how bad bad can get because the worst decade is not in the sample.

DIY investors commit the same error every time they:

• Use the S&P 500 1928-2025 record as “the stock market” while ignoring 34 other markets in the DMS database.
• Calibrate “normal” bond returns from the 1982-2020 falling-rate window.
• Calibrate “normal” growth-stock returns from the 2009-2021 zero-rate window.
• Use 2010s 60/40 outcomes to baseline retirement readiness.

Easy data bias is not the same as survivorship bias, though they rhyme. Survivorship bias is about who is in the sample. Easy data bias is about when the sample begins. The fix is the same: extend the window to include the periods you would rather forget.

U.S. Exceptionalism Is Real, but It Is Not a Retirement Plan

The 2026 UBS Yearbook reports that the United States now accounts for 62 percent of total world equity market value, up from a much smaller share in 1900 (UBS GIRY 2026 release, March 2026). Dimson points out the awkward feedback loop: the U.S. became the largest market by outperforming, and the outperformance is baked into a planning narrative that assumes future U.S. dominance as a given.

This is not an argument against owning U.S. stocks. A U.S. investor earns in dollars, spends in dollars, pays taxes in dollars, and rationally holds a U.S. tilt for that reason. The argument is narrower: building a retirement plan on the assumption that the next 30 years of U.S. returns will match the last 30 is a bet on one country’s exceptional century, not a plan. See The Case for Global Equity Diversification for the rotation history of which country has led each decade since 1970.

GDP Growth Is Not Stock-Market Return

Dimson devotes a long section of the interview to the common intuition that fast-growing economies should deliver fast-growing stock returns. The data says otherwise. Jay Ritter’s 2005 paper Economic Growth and Equity Returns reports the cross-country correlation between real equity returns and per-capita real GDP growth at 0.007 for 19 countries from 1970-2002 and 0.005 for 13 mostly emerging countries from 1988-2002. For the original Dimson-Marsh-Staunton 16-country sample from 1900 through 2002, the correlation is meaningfully negative.

Ritter identifies four mechanisms that break the growth-to-returns link:

1. Expected growth gets capitalized into prices, so the buyer of an already-richly-valued growth market pays for the future before it arrives.
2. New corporations and new share issuance dilute existing shareholders. Much of the growth accrues to firms that did not exist when an index investor bought in.
3. A growing share of national income goes to labor and consumers in a competitive economy, not to capital owners.
4. High capital expenditure to support growth lowers the cash available to return to shareholders.

This is the engine behind the most common emerging-markets disappointment. China’s real GDP growth averaged roughly 9 percent per year from 1992 through 2024, and Chinese equity investors earned a small fraction of that. The Stock Market Is Not the Economy covers the U.S.-versus-ACWI mismatch in more detail.

The Exciting Industry Is Rarely the Best Stock

The UBS 2026 release leads with a vivid contrast. In 1900, railroads were dominant U.S. listed equities. By 2025, they are less than 1 percent of U.S. market capitalization (UBS GIRY 2026 release). Despite that collapse in share, the DMS database shows long-run rail equity returns outperformed both the road-transport and airline sectors that replaced them in narrative excitement.

Dimson’s point is not “buy railroads.” The point is that the inputs to shareholder returns are not the same as the inputs to industry growth. Returns come from price paid, free cash flow received, dilution and buybacks, competitive entry, and reinvestment economics. The growth narrative drives prices up; whether the growth eventually compensates the buyer depends on the price they paid.

Technology sits on the other side of the same lesson. The UBS release notes that the broad tech sector survived the dot-com collapse and delivered superior multidecade returns, but only for investors who held through the 2000-2002 drawdown that wiped out the NASDAQ’s peak by more than 75 percent. The sector that delivered the highest long-run return was also the sector that punished bad entry-price discipline most severely.

Diversification Is Risk Control, Not a Return Promise

On the podcast, Dimson is careful with the diversification argument. International diversification cannot raise returns for every investor simultaneously, because every long position is someone else’s short. A U.S. investor who diversified abroad from 2010-2024 reduced risk and gave up return. A non-U.S. investor who diversified into the U.S. over the same window won the opposite side of the trade.

The case for global diversification is that the country whose century is next is not knowable in advance. UBS Yearbook 2026 finds that diversification remains valuable even though U.S. market concentration is at its highest level in 100 years and equity-bond correlations have risen. Most markets benefit from global rather than purely domestic equity investment when currency risk is hedged. See Global Revenue Is Not Diversification for the related argument that S&P 500 multinationals do not substitute for owning foreign-domiciled equities.

Factors Are Real, but They Need Patience and Humility

Dimson is not dismissive of factor investing. He acknowledges that size, value, momentum, and profitability have produced long-run premiums in the DMS data and elsewhere. He is cautious about how DIY investors should hold them. On the podcast he flags three practical realities:

• Long droughts are normal. Size and value have each gone more than a decade with negative or zero premium in some markets.
• Momentum has the strongest long-run record but the highest turnover, which makes its after-cost and after-tax return more fragile than its paper return.
• UBS Yearbook 2026 reports that 22 percent of country-decades have seen negative factor premiums across the full DMS sample. The premium is positive on average and over the full period; it is not positive everywhere all the time.

That last point is the one most DIY tilters underestimate. Tilting to small-cap value, for example, is rational with a 20-year horizon and tolerance for tracking error. It is painful with a 5-year horizon and a comparison habit. See Fama-French Factors and Small-Cap Value for the implementation details.

The Equity Risk Premium Is the Most Important Number You Cannot Know Precisely

Dimson’s direct planning advice on the podcast lands on the equity risk premium. Across two passes through the topic, he argues that:

• A historical equity premium near 3 percent is a defensible forward planning anchor for a globally diversified equity portfolio over long-run government bonds.
• 4 percent is on the optimistic edge of plausibility.
• Combined with a real long-run government bond return around 2 percent, that implies a real expected equity return near 5 percent as a base case.

Five percent real is materially lower than the 7 percent figure embedded in most U.S. retirement calculators, and the difference compounds. A 30-year accumulation at 7 percent real produces roughly 1.7 times the terminal wealth of the same plan at 5 percent real. That is the size of the planning error from using the wrong anchor. Do Stock Valuations Still Matter? covers the Damodaran implied ERP that crosswalks this from another direction.

Stress-Test Your Plan

The most useful Dimson move for a DIY investor is to stop treating the equity risk premium as a single number and start treating it as a range. The calculator below builds expected real portfolio returns from inputs the way Dimson does on the podcast: real bond return plus an ERP, minus fees and tax drag. Then it shows how much of your planned ending wealth depends on which ERP you assume.

Who This Matters For

Strong fit

• Long-horizon accumulators (10+ years to retirement) building a plan they want to survive a wide range of futures.
• DIY index investors who want an evidence-based reason to hold ex-U.S. equities beyond “recency.”
• Engineers and quants who already think in terms of dataset bias and parameter uncertainty.
• High-income earners whose human capital, employer stock, and home equity are already concentrated in one country, one industry, or one currency.

Poor fit

• Anyone investing money needed in less than 5 years. The DMS record is about long-run survivorship, not short-run timing.
• Anyone looking for tactical 6-to-24-month positioning. Dimson is explicit on the podcast that long-run history is the wrong tool for short-run signals.
• Anyone who wants individual stock picks. The framework is about how to size assumptions, not which security to own.
• Anyone who wants certainty. The DMS data shows wider outcome ranges than most investors plan for.

A DIY Investor Checklist

• Anchor planning returns on global, not U.S.-only, history.
• Use a real (inflation-adjusted) equity return around 5 percent as your base case. Stress-test at 2 percent and at 6 percent.
• Hold global equities at least proportional to global market capitalization, with home-country tilt only as a deliberate spending-currency hedge.
• Avoid country, sector, or theme overweights that depend on a growth narrative without checking what is already priced in.
• If you tilt to factors, commit to a 15-to-20-year horizon and accept that 1 country-decade in 5 has historically had a negative premium.
• Hold bonds and cash for spending needs and risk control, not because they maximize long-run wealth.

Frequently Asked Questions

Should I sell my U.S. index funds and buy global?

Not based on this post. Dimson’s argument is about how to size assumptions in a long-run plan, not about market timing. Selling a concentrated U.S. position to buy global today realizes embedded capital gains and bets that the rotation timing is now. The defensible action is to direct new contributions toward a more globally diversified allocation and let the home-country weight drift down over time.

Is 5 percent real too pessimistic given recent S&P 500 returns?

Recent S&P 500 returns are exactly the input Dimson is warning about. The U.S. compounded around 6.6 percent real from 1900-2025, but ex-U.S. equities compounded near 4.4 percent over a similar span. A globally diversified investor sits between those numbers. Five percent real is the central anchor, not the floor. The calculator above lets you see how your plan looks at higher and lower assumptions.

Does Dimson recommend factor investing or not?

He recommends it conditionally. The premiums exist in the DMS data, but 22 percent of country-decades have seen negative premiums and the highest-return factor (momentum) is also the highest-turnover and most cost-sensitive. The reasonable position is a small, low-cost tilt that you can hold for 15+ years without comparing to the cap-weighted index every quarter.

If GDP growth does not drive returns, why invest in emerging markets at all?

Diversification, not growth-chasing. Emerging markets have different drivers than developed markets, often different cycles, and lower correlations than implied by their growth headlines. Adding them to a portfolio reduces concentration risk in a single economic regime. The case is risk reduction, not return maximization.

How is easy data bias different from survivorship bias?

Survivorship bias is about which entities are in the dataset. Easy data bias is about when the dataset starts. A U.K. stock series that begins in 1948 has no survivorship bias (every listed firm is in the sample), but it has severe easy data bias because the worst decades for U.K. equities are pre-1948 and therefore excluded.

Where do I get the global equity exposure Dimson argues for?

Total-world equity index funds (VT, FZROX-equivalents, or a VTI+VXUS pair held at market-cap weights) are the simplest implementation. Holding only VTI or VOO concentrates you at 62 percent of the global market in a single country. Holding a small-cap-value tilt on top is a separate, optional layer.

Key Takeaways

• Optimism with conditions. Dimson’s long-run case for owning productive assets is real, but it requires diversification, low costs, patience, and realistic expected-return assumptions.
• Easy data bias is its own problem. A clean backtest that starts after the chaos has been edited out is not the same record as a global, century-long dataset that includes the chaos.
• Growth does not equal returns. Ritter (2005) finds the cross-country correlation between real GDP growth and real equity returns is roughly zero in recent decades and negative over the full 20th century.
• Diversification is risk control. Owning the world is a hedge against not knowing which country’s century is next. It is not a promise that every part of the portfolio wins together.
• The ERP is a range, not a number. A 3 percent equity risk premium and a 5 percent real equity return are defensible planning anchors. Plans that only work at 7 percent real are bets on the optimistic edge.
• Factors require time horizon and tolerance. 22 percent of country-decades have shown negative factor premiums. Tilting only pays for investors who can hold through the dry stretches.

Related Guides

• The Real Return of Stocks derives the 5 percent global real anchor from DMS, McQuarrie, and Jordà et al.
• Do 200 Years of Stock Returns Still Matter? covers regime dependence and survivorship corrections in the pre-1862 U.S. record.
• The Case for Global Equity Diversification shows decade-by-decade leadership rotation and Vanguard’s forward-return projections.
• The Stock Market Is Not the Economy explains why GDP growth and shareholder returns decouple.
• Do Stock Valuations Still Matter? decomposes forward returns into dividend yield, earnings growth, and multiple change.
• Fama-French Factors covers the size, value, profitability, and momentum framework Dimson works inside.
• Stocks Are Always Risky covers the deep-risk counterexamples (Russia 1917, Japan post-1989) that long horizons do not repeal.
• Safe Withdrawal Rate shows how the ERP assumption flows into sustainable retirement spending.

Sources

• Rational Reminder Episode 408, “Elroy Dimson: Investing & Optimism”, May 7, 2026.
• UBS Global Investment Returns Yearbook 2026. Release and public summary PDF.
• Jay R. Ritter, “Economic Growth and Equity Returns,” Pacific-Basin Finance Journal, 2005.