Invest1 publisher2 min readPublished
Full reversion from a CAPE of 41 would cost the S&P 500 8.4% a year for a decade
Robert Shiller's cyclically adjusted multiple is in the top 1% of readings since 1881. It caps how much of the next decade's return can come from the multiple. It does not date a top.
The Investor · Invest desk
What happened
- Robert Shiller's cyclically adjusted price/earnings ratio for the S&P 500 has climbed to about 41.1, the highest reading since the peak of the dot-com bubble.
- The long-term average of the measure since the 1880s is around 17, which puts the index at more than twice its historical norm for a dollar of normalized earnings.
- The all-time high in the series is 44.2, set in December 1999, immediately before a crash that erased trillions in market value over the following two years.
- The ratio has stayed above 40 continuously since May 2026, a plateau Crypto Briefing says has no real precedent outside the brief window around 1999 and 2000.
- The Buffett indicator, total US market capitalization against GDP, exceeded 237% in September 2026, against the 200% level generally read as significant overvaluation.
Compiled by The InvestorSomething wrong?How this is made
Why it matters
- constraint Carrying the last decade's equity returns forward for another ten years now requires the multiple to stay among the most expensive readings the series has recorded, so multiple expansion belongs in the aggressive case.
- decision Anyone modelling a ten-year equity return has to state a terminal multiple out loud, because that single assumption sets the sign of the price return before any earnings estimate is entered.
- contradiction The same article uses the ratio as a crash precedent and as a ten-year return forecaster, and only the second use has the correlation behind it.
- exposure No one in the material puts a figure on how many points of CAPE the accounting, buyback and composition adjustments are worth. That unpriced number decides how much of the drag is real.
Invert 41.1 and a dollar of index buys about 2.4 cents of the ten-year average of inflation-adjusted earnings that sits in the Shiller denominator [4][1]. Invert the historical average and the same dollar buys 5.9 cents [2].
Reversion is the expensive part. Going from 41.1 to 17 takes 58.6% off the multiple [4], and spread over ten years that is 8.4% a year off the price [5]. For the index to come out flat in real terms across that decade, ten-year average real earnings would have to compound at about 9.2% a year [6].
A partial reversion is the more interesting version. If the multiple settles at 30, the bottom of the band Crypto Briefing associates with weak decade-ahead returns, the drag is 3.1% a year [10].
Upward, the room is 7.5%. A multiple that gains more than that from here sets an all-time high [3].
The ratio first went through 40 in January 1999 and the record came eleven months later [6][9]. Crypto Briefing's summary of the long-run evidence is that starting valuations above 30 to 40 have correlated with weak equity returns over the following ten years [11].
The objections to the series are real, and the article lists them: changed accounting standards, buybacks in place of dividends, and index composition tilting toward higher-margin technology companies, each of which argues for a structurally higher modern reading [10]. Crypto Briefing grants those arguments merit and says they do not fully explain a reading of 41 against an average of 17 [10]. On the second gauge, the publication reports that Warren Buffett described a version of the market-capitalisation-to-GDP measure as probably the best single measure of where valuations stand at any given moment [9].
In my view the usable conclusion is narrow. Out of roughly 1,700 monthly observations since 1881, the top 1% holds at most about 17 of them, and the current reading is in that group [3][8]. I would be wrong on the drag if ten-year average real earnings compound near 9% a year for the next decade, or if the fair multiple for a technology-weighted index genuinely is closer to 40 than to 17; the second cannot be settled inside ten years, and the first turns up in reported earnings every quarter.
What to watch
- Fourth-quarter earnings entering the ten-year average: a stronger denominator lowers the multiple with no help from falling prices.
- The Buffett indicator's path down from 237% while the CAPE holds above 40 would split the two gauges and weaken the confluence argument.
- Any attempt to quantify the composition adjustment: the counter-thesis needs a defensible fair-value CAPE for a technology-weighted index.