Roger Ibbotson built the first modern index of US stock and bond returns 50 years ago, measuring 1926 through 1975 and forecasting out to the year 2000.
Almost every long-run return study since has started at 1926, because that is where the clean US data starts. The book Ibbotson has just brought out pushes the record back to a French water mill with share quotes from 1372, and it finds that in the 130 years before 1926 bonds did far better against stocks than the modern era taught investors to expect.
"So this seems to be a part of human nature. It's not just an invention by Americans in the 20th century."
Ibbotson is professor emeritus of finance at Yale School of Management and chairman of Zebra Capital; he sold Ibbotson Associates to Morningstar in 2006, and the CRSP data behind this book reached the project through that sale. Siegel is director of research emeritus at the CFA Institute Research Foundation, which published the book, and its executive editor.
I listened to the full interview so you can skip it. 35 minutes of audio, 16 minutes of reading.
Here are the 12 insights that matter.
👤 Guests: Roger Ibbotson, Professor Emeritus of Finance at Yale School of Management, Chairman of Zebra Capital and the book's main author, and Laurence Siegel, Director of Research Emeritus at the CFA Institute Research Foundation, the book's executive editor and a co-host of this podcast
🎙️ Host: Lotta Moberg, who hosts the Financial Thought Exchange podcast for the CFA Institute Research Foundation
📰 Published: 10 September 2026, on YouTube
🔴 YouTube | ⏱️ 35 min | ✅ Time saved: 19 min
Key Takeaways
From 1793 to 1926 stocks did about as well as they did in the 20th century, but bonds did much better, so the equity risk premium was low and in some periods zero
The reason was that the US went from emerging market to developed market and yields fell, handing bondholders a large capital gain
The small cap premium has barely existed for 50 years, and the reason is that it got discovered
Siegel: small caps were once called secondaries, were repriced upward as the effect became known, and now behave like any other stock
Ibbotson said on television that SpaceX was a great short-term trade and a terrible long-term investment, because 95% of the stock was held by people who wanted to sell
It was oversubscribed and shot up from 135 at the listing; he put it at about 110 now
The US record is the winning country's record, so the book refuses to extrapolate it
The forecast out to 2050 pairs today's yield curve and US equity risk with the lower global equity risk premium from the Dimson-Marsh-Staunton data
Cap-weighted indexes are now the least diversified version of themselves in a century
Ibbotson will not call that bad, but says it carries risks a broader market does not
The Dow is price-weighted because its compilers had no calculators
Ibbotson: price weighting "really doesn't make any economic sense", and nothing but the Dow still uses it
Index-level volatility has fallen, and the new Fed chair thinks that is a problem
Ibbotson said Kevin Warsh wants more volatility because the Fed has over-smoothed the market by talking about the future
1. The 100-Year Index Project
Ibbotson opened on the history. Fifty years ago he and Rex Sinquefield built the first indexes of the US stock and bond markets, measuring the 50 years from 1926 to 1975 and making forecasts out to 2000. He then ran Ibbotson Associates, which produced the yearbooks and the index series for years.
He sold Ibbotson Associates to Morningstar in 2006; Morningstar later bought CRSP, which is the source of the data in the new book, so the project has the rights to produce the series
Rebuilding a 100-year series today is as much a licensing exercise as a data exercise — all capital-markets data now belongs to somebody, and the work involves the deals and licenses as much as the arithmetic
The book runs on 100 years of monthly US data, but Ibbotson said he did not want to look only at that data, which is why he brought in co-authors covering other centuries, other countries and other asset classes
Siegel's case for why the exercise matters at all is compression: an index packs an enormous amount of information into a single number, which is what lets anyone answer the question of how the market did last year
2. Prices Back To 1372
The finding Ibbotson called one of the most interesting to come out of the co-authors is how far the record actually reaches.
Will Goetzmann, a co-author, studied a water mill in France that has stock quotes back to 1372, and also found evidence of capital markets with prices in antiquity, in Ancient Rome
Ibbotson's conclusion from that: "So this seems to be a part of human nature. It's not just an invention by Americans in the 20th century."
The practical consequence for the book is that it does not stop at the US monthly series — it carries chapters on the pre-1926 US market and on global returns as well
3. Why Cap Weighting Won
Moberg asked what a reader should know about index weighting before anything else. Ibbotson started with the Dow Jones Index, which is price-weighted — you add up the share prices.
"They didn't have calculators. They had to add things up by hand, so they just price-weighted it." The method was chosen for ease of calculation, not for economics
Asked by Moberg what actually separates a price-weighted index from a cap-weighted one, Ibbotson walked through it: price weighting looks only at the price per share, while capitalization is price per share times the number of shares. A company can therefore change its Dow weight by changing its share count — he offered a million-dollar company held either as a million shares worth $1 or as one share worth a million, and said the Dow would reward the second
His verdict on it is flat: price weighting "really doesn't make any economic sense", and nothing but the Dow still uses it as a holdover
A cap-weighted index multiplies every company's price by its share count, so the weights sum to 100% and large companies carry large weights — which is why Ibbotson calls it the measure of what has happened across the whole market
Equal weighting is common, and Moberg's point was that it makes the index's inclusion and exclusion decisions matter more than cap weighting does, because a newly included company arrives at full weight rather than at a small one
Morningstar's and CRSP's Vanguard indices define large cap as roughly 80 to 85% of total capitalization, which means the number of stocks floats. The book instead cuts by count — the 300 largest companies over the last 50 years, which is how S&P and Wilshire do it
Counting stops working the further back you go. There were only about 500 public companies in the data set in the 1920s, so measuring large caps in that era needs a smaller count; CRSP uses deciles instead
4. Why Bond Returns Are Harder
Moberg put it to both of them that a stock return is obvious — somebody puts a new price on it — and asked whether bonds are as straightforward. Ibbotson's answer was two words long: "Much more complicated and much more mathematical."
The mechanics are the reason. A bond index has to start from the Treasury yield curve, and the raw material for a yield curve is forward interest rates, which have to be estimated. From those forward rates any curve can be built — most commonly a par bond curve, where every bond is priced at 100. No Treasury actually trades at 100, so the curve is synthetic by construction, and every other bond in the market is then priced off it.
Ibbotson worked the techniques out with Tom Coleman, a co-author on the bond side, starting in the 1980s — about 40 years on the same problem. He said they will be interviewing Coleman on the show
Asked whether there is guesswork involved, he said the opposite: bonds are priced more carefully than equities, because every bond can be priced from what another bond is priced at
The split he drew is that bond analysis is mathematical and equity analysis is statistical, because so much of a stock's return is idiosyncratic rather than common
"The thing is that you don't know what a company is gonna do when you buy the stock. It could prosper, it could fail, it could purchase other companies."
"Bonds have a schedule of cash flows that the bond issuer is supposed to pay, and so you can analyze it much more mathematically and come up with much more solid answers."
5. The Winning-Country Bias
The book's international material comes from Dimson, Marsh and Staunton, who measured 125 years of returns across what Ibbotson put at about 20 countries. It is annual data rather than the monthly US series, and the book carries a summary of their results.
Moberg raised what that long a window does to the sample: Russia, China and Japan all left the market at some point, Japan came back, and Austria and Hungary did not.
Ibbotson's answer is that the US is the survivor and the winner at once — "the US has a survival bias because it not only survived all these wars and so forth, but it also prospered and so it turns out to be the winning country over the last 100 years"
So the forecast in the book's last chapter deliberately does not extrapolate the US record. It pairs today's yield curve and the risk of the US stock market with the lower equity risk premium from the global data, and runs out to 2050
Ibbotson said the method is mostly to take the past and extrapolate it, adjusted for inflation and the yield curve — but that a forward expectation built on the US alone is probably too high, and the international results are the better input
He also flagged that the equity risk premium is simply the difference between the stock return and the bond return, which is the number the survival correction is aimed at
6. Bonds Won Before 1926
Siegel brought in the pre-1926 work, and it is the finding that most changes the picture.
"One of Ed McQuarrie's most interesting findings is that if you go back before 1926, let's say 1793 to 1926, the stock market did about as well as it did in the 20th century."
The bond market over that stretch did much better, so the equity risk premium was low — and over some periods it was zero
Siegel's explanation: "it's because the United States went from being an emerging market to a developed market in that period, and yields came down, which meant you got a huge capital gain on those bonds"
Ibbotson said he and Goetzmann had looked at the pre-1926 US data themselves and had not found all of it. McQuarrie, working with what the internet now makes findable, filled in the gaps and got back to the founding of the New York Stock Exchange in the 1790s
The book uses McQuarrie's data set rather than Ibbotson's own — "If somebody does something better and they have better techniques, you wanna use that, basically"
7. Measurement Traps
Moberg pressed on whether the US sample has survivorship problems of its own, inside the country rather than across countries. Ibbotson said there are several, that they run in both directions, and that the job is correcting them rather than avoiding them.
When a company is bought out, the series has to capture the final price rather than dropping the company at its last quote
"You wanna make sure that when there's a bankruptcy, that you capture what's left of the company, and you include the negative return that's associated with the bankruptcy even after it's kind of quoted anymore."
The problems compound the further back the data goes — the 1800s, the 1700s, the 1600s. "We know less and less and less about that data as we go back in time."
He still argued the measurement is worth doing at that distance, on the grounds that knowing something about Roman-era prices is valuable even when the mismeasurement is certain
8. The Emerging-Market Trap
This section is Moberg's own argument, and Siegel's response to it.
Her point: run the numbers back 100 years and emerging markets have not returned much, and the standard conclusion is that they are a poor investment carrying higher risk. But nobody investing in 1900 knew which markets would be labeled emerging a century later, and the label exists precisely because those markets did not grow — which makes the comparison circular
Siegel agreed that investors bias their holdings toward their home market, and offered the counter-example that runs the other way: Argentina in 1900, which he said probably looked to investors then much as the US market did
"It had tremendous prospects, people thought, and that one did not turn out at all."
9. SpaceX And The Free Float
Moberg raised SpaceX's listing as the event that taught a wide audience the difference between market capitalization and free-float-adjusted market capitalization, and asked which an index should use.
Ibbotson said the free float is clearly the right measure for SpaceX. The offering was heavily oversubscribed and shot up from 135 right at the onset; he put the stock at about 110 now, which he still called high on any normal fundamental metric
"I said on TV that it had really poor long-term prospects because first of all, 95% of the stock was still held by other people who wanted to sell, and only 5% of it was newly issued here." The 95% included venture capital and private equity holders of long standing who wanted out
"So I definitely recommended that this was a great short-term investment with the oversubscription but a terrible long-term investment."
On index construction his answer is mechanical rather than valuation-driven: "If you had an index which used the full capitalization of SpaceX instead of free float, the index funds would be forced to buy much more than they can buy." At full capitalization SpaceX would be one of the largest stocks in the index and would overwhelm it
The book's own indexes are not free-float adjusted, and cannot be. Ibbotson said the free-float numbers only exist from roughly 15 years ago; the back data does not carry them, so the book measures everything rather than the tradable slice
10. Small Cap's Premium Faded
Moberg framed the small cap premium as the one factor its believers will say has always worked, and noted the book shows it diminishing.
Ibbotson's read of the chart: small caps were hammered in the Depression, then had a tremendous run from the 1970s into 1980, after the 1973-74 recession
"In the last 50 years, there really hasn't been much of a small cap premium, actually." He added that small caps did great this year, but that the long-run payoff has not been there
Siegel's explanation is that the premium was arbitraged by being discovered. Small caps were once called secondaries and most investors did not buy them because they wanted blue chips; as index funds spread and the small cap effect became better known, they were repriced upward, and that repricing is where the historical returns came from
"But after they became more fairly priced, in other words, more expensive, then the forward-looking returns on them became more like any other stock."
Moberg asked whether the people still recommending small caps at conferences have simply not updated their data since the 1980s. Siegel called that too negative and said he still believes the premium exists — there is less information on small companies, they are less liquid and harder to buy, so there are reasons it should be there
"So there are reasons why there should be a small cap premium, but there hasn't been lately."
11. Concentration Is Near Peak
Ibbotson said cap-weighted indexes, "the gold standard of being diversified", are now less diversified than they were, because a few companies carry so much of the weight
He attributed it to the digital revolution: the companies concerned are primarily technology, or at least companies with a technological edge, however they happen to be labeled — media, or something else
The market is at almost historically high concentration on the book's measure. He noted the number is harder to compute across the full 100 years because there were fewer companies in the early period
There is precedent, and it was benign. In the 1950s General Motors, AT&T, Ford and some of the big chemical companies had large capitalizations relative to everything else, and Ibbotson said that worked out pretty well — all of them have since declined in importance because others grew
"But it's not a foregone conclusion that having a concentrated market is bad. But there are some risks to it that don't exist when the market is more diversified."
12. Warsh Wants Volatility
Moberg's last observation was that index-level volatility has declined over the period, which she said she would have expected to move the other way as diversification fell.
Ibbotson confirmed the direction: volatility was tremendous during the Depression, turned in the middle of World War II as the US started to win the war, and has if anything drifted lower since
He then brought in the Federal Reserve chair's view, which is that the market should be more volatile than it is. Ibbotson said Kevin Warsh does not want the Fed talking its way into the future
"He wants to be able to react to things that happen and not have artificially low volatility."
"Because he thinks it's being over-smoothed by the way the Fed has handled things."
Ibbotson's own gloss is that the smoothing may be hiding rather than removing risk: so much information about the future is constantly released that measured volatility is artificially low, while the large discrete events still happen
He tied that to Knightian uncertainty — Frank Knight's case of not knowing the probability distribution at all — and to 1987. "Like the crash of '87 was beyond what could reasonably happen from a probabilistic perspective, and yet it didn't have much impact on the long-term returns."
Bonus Insights
Ibbotson's interest in the SpaceX question is not incidental: his dissertation was originally on initial public offerings, and he said he was investing in IPOs as a teenager
Moberg pointed listeners to the show's two earlier conversations with Rob Arnott on index construction and fundamental indexing, where the effects of index inclusion and exclusion were covered
Siegel's case for the book against Ibbotson's earlier work: "But as we get into the other asset classes, the other centuries, the other countries, you learn more fully about the capital markets." He called gathering many researchers in one place the book's marginal contribution over Stocks, Bonds, Bills, and Inflation
He also said the value is partly in making the data accessible and telling readers what data exists at all — the book cannot contain all of it, but it synthesizes and analyzes what is there
Moberg noted the first of the two conversations covers how these returns connect to GDP growth, and said more episodes with the book's individual co-authors are planned
The bottom line both men land on is that a century of US returns is the record of the one country that won, and that the further back and further out the data goes — to 1793, to 1372, to 20 other countries — the lower the equity premium an investor should expect to be paid.
Products, Companies & Tools Mentioned
Morningstar and CRSP (Morningstar bought Ibbotson Associates in 2006 and later CRSP, which is the source of the 100 years of data behind the book)
SpaceX (Ibbotson said on television that it was a great short-term trade and a terrible long-term investment, because 95% of the stock was held by sellers and only 5% was newly issued)
Dow Jones Industrial Average (Price-weighted because its compilers had no calculators; Ibbotson says the method makes no economic sense and survives only as a holdover)
Vanguard (Its Morningstar- and CRSP-built large cap index is defined as roughly 80 to 85% of total market capitalization, so the stock count floats)
S&P Dow Jones Indices and Wilshire (Cut the market by stock count, which is the method the book uses — the 300 largest companies over the last 50 years)
Zebra Capital (Ibbotson's firm; he chairs it)
General Motors, AT&T, Ford (The 1950s concentration comparison: all had large capitalizations relative to the rest of the market, and Ibbotson says that worked out pretty well)
Books & Resources Mentioned
Exponential Wealth: Centuries of Stock and Bond Returns – Roger G. Ibbotson, main author, and Laurence B. Siegel, executive editor (The CFA Institute Research Foundation publication the conversation is about, with 100 years of US data plus chapters on earlier centuries and other countries)
Stocks, Bonds, Bills, and Inflation – Roger Ibbotson and Rex Sinquefield (The 50-year-old series the new book succeeds; the original measured 1926 to 1975 and forecast to 2000)
Global Investment Returns Yearbook – Elroy Dimson, Paul Marsh and Mike Staunton (125 years of annual returns across about 20 countries, and the source of the lower equity risk premium the book's 2050 forecast uses)
Index Construction and Fundamental Indexing with Rob Arnott (The earlier Financial Thought Exchange conversation Moberg pointed listeners to on other ways to weight an index)
CFA Institute Research Foundation (The publisher; Moberg read out its address on air)
If this was worth your time, send it to someone who has to have a view on this.
Get the latest market chatter as it happens: