Stocks for the Long Run? New Evidence, Old Debates

For decades, Jeremy Siegel’s “Stocks for the Long Run” thesis, that equities reliably outperform other assets over long time horizons, has served as a foundational pillar of modern finance. His famous chart, showing real (inflation-adjusted) returns for stocks, bonds, bills, gold, and cash stretching back to 1802, became a visual shorthand for the preeminence of equities. But what if the story supported by these data is less definitive than it appears? And what if the data themselves are not representative of what really happened?

In this CFA Institute Research Foundation brief, “Stocks for the Long Run? New Evidence, Old Debates,” author Paul McCaffrey, Principal, Paul McCaffrey Enterprises, revisits Siegel’s thesis through the lens of Edward McQuarrie’s scholarship, which gathers and interprets a rich history of nineteenth-century US stock and bond returns. McQuarrie’s findings — published in the Financial Analysts Journal — suggest that for much of the 1800s and early 1900s, stocks did not consistently outperform bonds. In fact, returns on the two asset classes were often roughly the same.

This revisionism prompted a conversation among financial thinkers, including McQuarrie, Jeremy Siegel, Rob Arnott, Hendrik Bessembinder, Elroy Dimson, Roger Ibbotson, and Laurence Siegel, who contributed the foreword and afterword to this brief. Their ideas, first discussed in a 2024 CFA Institute podcast and continued in private correspondence, form the basis of this brief.

But the brief is not merely a reassessment of the broader “Stocks for the Long Run” thesis. It asks more fundamental questions: How are financial histories constructed? What flaws might they have? What assumptions are embedded in the data on which we base our portfolio allocation and investment decisions? And how do simplified investment narratives and stylized charts become financial orthodoxy?

McQuarrie’s critique highlights the potential flaws of long-term datasets. Jeremy Siegel filled historical gaps with estimates — such as his 6.4% dividend yield assumption applied to equities in the early nineteenth century — whereas McQuarrie used exact dividend data that he collected using original records from early exchanges and brokerages. The result is a very different picture of nineteenth-century equity performance—one that casts doubt on the constancy of the equity risk premium.

Stocks clearly dominate in this rendition of the past. But McQuarrie’s new data show a lower return for stocks in the first half of the period studied. He demonstrated that stocks decisively outperformed bonds only in a few subperiods, a finding that casts doubt on Siegel’s thesis.

In the end, this brief urges readers to treat investment narratives with caution. Behind every powerful chart lies a web of assumptions, and behind every average return lies a wide range of outcomes. Financial history, properly studied, is less a road map than a tool for asking better questions and conducting deeper analysis.