# The Math Is Simple — So Why Isn't Everyone Rich? Notes on Roger Ibbotson

> Listening notes on The Meb Faber Show #651 with Roger Ibbotson: a century of stock and bond returns, human capital and age, popularity-based pricing, and what he found when he graded his own fifty-year-old forecast. Educational commentary, not investment advice; no stock recommendations or price targets.

Published: 2026-09-18
Locale: en
Tags: compounding, asset allocation, long-term investing, human capital, valuation


![A university office at dusk, a long roll of graph paper unfurling across the desk and down the floor into a receding hallway, the near curve lit by a desk lamp and the far end fading into shadow](/covers/mebfaber-2026-09-18-roger-ibbotson-why-isn-t-everyone-rich-when-the-ma-cover.png)

> Remember that money is of a prolific generating nature. Money can beget money, and its offspring can beget more.

—— Benjamin Franklin, *Advice to a Young Tradesman* (1748)

Franklin wrote that in 1748. Two hundred seventy-eight years later the arithmetic hasn't gotten harder and the answer hasn't gotten smaller. The number of people who know it has grown by orders of magnitude. The number who pull it off has not. This episode spends an hour on the gap in between.

## What this episode covers

The Meb Faber Show, 18 September 2026, episode 651, with Roger Ibbotson — forty years teaching finance at Yale, founder of Ibbotson Associates, chairman of Zebra Capital. When anyone in this industry quotes long-run asset class returns starting from 1926, they are quoting his dataset. That chart with five lines on a log scale — large caps, small caps, bonds, bills, inflation — first appeared in 1974 and still shows up in decks today. His new book, *Exponential Wealth*, is out from the CFA Institute.

Meb opens with a question that has a little mischief in it: one dollar turns into more than fourteen thousand, so save a bit, invest a bit, leave it alone — a plan a child could execute. Why isn't everyone rich? Ibbotson doesn't dodge. He takes the conditions that number depends on and lays them out one at a time.

## The main points

**One: that 14,751x comes with three brutal conditions.** A dollar invested in large caps in 1926 grew to $14,751, about 10.1% a year. That assumes no taxes, no fees, no transaction costs, and every dividend reinvested. Ibbotson's first explanation is blunt: most people don't reinvest, they consume. Take a slice out along the way and the curve continues on a lower track. He adds that there is no alpha in any of this — you're buying the entire market, not picking winners. The hard part sits in not touching it.

![Two rising curves start from the same point; the upper one climbs steeply the whole way, while the lower one is broken mid-course by a downward gap and then continues growing on a lower track, leaving a wide gap in height between them at the right edge.](/figures/reinvest-or-withdraw-en.svg)

**Two: after inflation, 820x.** Prices rose roughly 18-fold over the century, so that 14,751x converts to about 820x in purchasing power. Their forecasts and reports always carry both nominal and real, because real is the number that buys groceries.

![A very tall bar stands for the nominal figure of more than fourteen thousand times, while only a small slice at its base is shaded to show the 820 times left after prices are taken out, and the difference in height between the two is enormous.](/figures/nominal-versus-real-en.svg)

**Three: crashes and booms carry no forecasting power by themselves.** After a crash you don't know whether it continues or rebounds; after a boom, same problem in the other direction. Knowing which regime you're in doesn't tell you what comes next. His example is 1987: a 20% drop in a single day, a blip you can barely find on the century-long log chart, and a slightly positive total return for the full year.

**Four: your equity weight is a function of human capital.** Ibbotson has studied human capital for decades, and his recommendation for young people is 100% stocks in the investment portfolio — the reasoning isn't that the young tolerate drawdowns better, it's that most of their wealth still sits in future earnings, with financial capital a small slice. As you age, the time available to recover shrinks and the human capital depletes, so you should get more risk averse. He gets wry about himself here: at my age there isn't much human capital left, financial capital is basically all I've got, and since I was mostly in the stock market, I'm not complaining.

![Two crossing wedges: human capital, standing for future wages, narrows from the upper left down to the lower right, while financial assets grow from the lower left up to the upper right, and the two cross in midlife.](/figures/human-capital-crossing-en.svg)

**Five: the bond mountain, and what "diversified" actually promises.** Yields bottomed around 1940, climbed to double digits by 1980 on the back of inflation, and after Volcker began a forty-year descent that ran until roughly 2022. A bond's return decomposes into yield minus duration times the change in yield: a ten-year duration long bond loses 10% when yields rise one point. So 1940 to 1980 delivered coupons swamped by capital losses; 1980 onward delivered high coupons plus capital gains, both at once. Then Meb raises correlations. Small caps and long bonds correlate near zero over the full history — but zero is an average. The relationship runs strongly negative for a stretch, strongly positive for another stretch, and each stretch lasts long enough that you can tell which one you're in. 2022 was one of the positive stretches, and both sides fell together.

**Six: popularity is where the discount comes from.** This is the model at the centre of the book. Risk is unpopular, so risky assets get priced lower and carry higher expected returns — that's what CAPM is describing. Liquidity is popular, so illiquid assets have to offer more expected return to clear. Push one layer further and you get reputation: two companies with identical expected cash flows, the one with the good brand trades higher, the one with the poor reputation trades lower with a higher expected return. He uses a bond as the analogy — yield up, price down — and says equities work the same way underneath. And when something gets popular past the point of sustainability, pricing starts to break.

![A mountain of yields: climbing from the 1940 low to the 1980 peak and sliding down to the 2022 low, with the up slope in one color and the down slope in another, and a small box in the upper right saying that at a duration of ten years a one point rise in yields costs ten percent.](/figures/bond-yield-mountain-en.svg)

**Seven: he graded his own 1976 forecast, and he marks the US down on purpose.** Fifty years of out-of-sample data now exist, so he redrew the old chart in the same font and layout to check. Through 2000 the nominal numbers tracked closely enough that he went on television and took credit. After that, nominal ran high (inflation fell) and real ran low, with the outcome landing in the middle of the distribution. The book's forecast to 2050 is 7% nominal, 5.6% real — and it does not simply extrapolate the US equity risk premium. The Dimson, Marsh and Staunton chapter covers 125 years of global data and shows the US as close to the best-performing market in the sample. Argentina looked wonderful in 1900. Markets on the losing side of the world wars got broken outright. So he subtracts the roughly 1.5 points between the global average and the US, on the grounds that you can't assume the US wins the next twenty-five years too.

**Eight: a 1% dividend yield is not a reason to mourn.** He wrote a paper with Philip Straub at Morningstar on buybacks. Add buybacks to dividends and total cash payout runs around 4%, a figure that has held roughly steady for a couple of centuries. What changed since the 1980s is the form — buybacks are more tax efficient for shareholders, letting you realise the cash when you want it and pay capital gains, rather than being taxed on the company's schedule.

![Two companies with identical cash flows sit side by side: the well-liked one has a tall price bar and a short expected-return arrow, while the disliked one has a short price bar and a long expected-return arrow.](/figures/popularity-is-the-discount-en.svg)

## Going further

### "I ran the numbers. Why doesn't my account look like the spreadsheet?"

This is the frustration an episode like this tends to produce. You type 10% into a cell, compound it thirty years, get a cheerful number, then open your statement and find a different story.

Ibbotson's first-layer answer is direct: most people consume it. That deserves unpacking, because it sounds like a lecture about discipline when the thing actually operating is how sensitive exponentials are to interruption. The bulk of the money arrives late in the curve, so any withdrawal costs you more than the amount withdrawn — it costs that amount's ability to keep replicating itself through every remaining year. Ten thousand pulled out at twenty-five and ten thousand pulled out at fifty-five are the same cell in the spreadsheet and two orders of magnitude apart in the outcome.

![Three growth curves start from the same point: one runs unbroken, and the other two are each interrupted by a gap of the same size, one early and one late, with the early one ending far lower among the three endpoints on the right.](/figures/when-the-money-leaves-en.svg)

The second layer is the taxes, fees and transaction costs he mentions in passing. That chart is the zero-cost version, a physical ceiling rather than a number anyone collects. What you collect is the ceiling minus friction, and friction is one of the few inputs you steer directly.

The third layer is the uncomfortable one: for most people the interruption is not a decision, it's life arriving. Job loss, a move, a medical bill, a house. This loops straight back to his human capital argument — whether your equity position survives a bad stretch depends on whether your earnings stream broke in the same month. So the first line of an allocation plan should read "how many months of cash let me avoid selling when I get hit." Leave that line blank and the 10% underneath it is decoration.

### "CAPE is screaming expensive again. Should I get out?"

The way Ibbotson handles this beats any conclusion he could have stated: he tells you what happened to a person. Shiller's office is down the same hallway. *Irrational Exuberance* came out in early 2000 and the tech bubble broke right after, timing that looks staged. Then Ibbotson supplies the second half — Shiller himself, acting on his own CAPE, had exited the stock market several years earlier and missed the late 1990s entirely. His verdict: the publication lag worked in his favour that particular time.

There's a usable reading in that. Valuation is a water level, not a clock. It tells you which way the distribution of ten-year returns tilts for someone buying today; it says nothing about when that tilt gets paid out. "When does it resolve" and "can you stay solvent until it resolves" are separate questions, and the second one is answered by your cash flow and your holding period. The most honest line in the episode may be this one: CAPE says the market is wildly overvalued today, and it said the same thing in the mid-nineties.

![On the left are two bell-shaped distributions with the high-valuation one shifted to the left as a whole; on the right is a clock face that has markings but no hands.](/figures/a-level-not-a-clock-en.svg)

So if you're sitting on a valuation signal and hesitating, the move with better odds is to rewrite the question from "should I liquidate" into "how long is my holding period, and if this drags five years will something force me to sell." The first asks you to guess a date. The second asks you to inventory your own situation.

### "I diversified. Why did everything fall together in 2022?"

The answer hides inside the definition of a correlation, and the report usually prints only the single number. Small caps and long bonds correlate near zero across the full history — a midpoint averaged from a very positive era and a very negative one. No trading day ever runs at the average. Ibbotson goes further and says there's some predictability here: when they're running negatively correlated they run that way for a while, and the same holds in the other direction.

![A curve swings between positive and negative for long stretches, a dashed line through the center marks a long-run average near zero, the upper band is labeled falling together and the lower band offsetting each other, and a dot at the right edge marks 2022.](/figures/correlation-swings-around-zero-en.svg)

What that means in practice is that when you buy diversification, you buy a long-run property, and you live through its extremes in particular years. If your plan needs to hold every single year, the average can't help you. If your plan only needs to hold across twenty, it's enough. Which camp you're in depends on when you need the money, not on how you rate your own tolerance for volatility.

Worth keeping too is his passage on tails: a lognormal distribution isn't sufficient to describe the shape of equity returns, you need skewness and kurtosis on top, and fat tails mean things absent from the dataset will happen — up to and including the bankruptcy of a whole economy. A hundred years sounds long in statistics. In human history it isn't.

## Where to look next

- The Meb Faber Show, episode 651, 18 September 2026, with Roger Ibbotson.
- Roger Ibbotson et al., *Exponential Wealth: A Century and More of Stock and Bond Returns*, CFA Institute. Includes a chapter by Elroy Dimson, Paul Marsh and Mike Staunton on 125 years of global market data.
- Robert Shiller, *Irrational Exuberance*, first edition 2000 — the origin of CAPE and the story about its timing.
- The Ibbotson and Philip Straub paper on buybacks and total payout yield.
- His early work on IPO underpricing and hot issue markets — the episode notes that 1970s IPOs rose about 15% on the first day and underperformed afterward, a structure that persists today, except the companies now queuing up are already giants before they list.

## The one thing to take away

One idea stayed with me: **the cost of an interruption isn't the amount withdrawn, it's that every year afterward restarts from a lower base.**

This doesn't only happen in accounts. Practising an instrument, running, writing a paragraph each week, phoning your parents each month — same structure. The day you stop feels like the loss of one day, while every day after it builds from one step lower. I've fallen off several of these, and every time I told myself the reason was that I'd been busy.

Here's something I tried, offered in case it's useful. Pick one small thing you kept up for more than three months and then dropped. Dig through your calendar or your photos and find the last day it happened, then write down what actually occurred — down to the grain of what time it was, where you were, what was in your hands. "Too busy" doesn't count. What usually surfaces is something small and specific: you were at the office until nine thirty, the kid had a fever, it was raining and you couldn't face the door. Then lower the restart threshold until that specific thing can't block it — the five-minute version, the do-it-in-the-living-room version, the three-sentence version.