CAPE Ratio Explained: When Stocks Are Truly Expensive

TL;DR: The CAPE ratio — Robert Shiller's cyclically adjusted price-earnings ratio — divides the S&P 500's price by the average of its last ten years of inflation-adjusted earnings. The point is to strip out the business cycle. At 41.43 in June 2026, CAPE is more than double its long-term mean of 17.4 and within one point of its 1999 all-time high. That is a statement about valuation, not timing.

What CAPE actually measures

Robert Shiller and John Campbell introduced the cyclically adjusted price-earnings ratio in a 1988 paper. The motivation was straightforward: a normal price-to-earnings ratio looks rich in recessions (when earnings collapse) and cheap at the top (when earnings are unsustainably high). A smoothed earnings denominator fixes both problems.

The formula:

CAPE = Real S&P 500 price ÷ Average real earnings of the prior 10 years

Both the numerator and the ten years of trailing earnings are converted to today's dollars using the CPI. Shiller's monthly series goes back to 1881 and is published on his Yale data page as ie_data.xls.

A worked example

Suppose the S&P 500 is at 6,000 today, and the index's real (inflation-adjusted) earnings over the prior ten years averaged $145 per share. Then CAPE = 6,000 ÷ 145 ≈ 41.4. That is roughly where the index sits in June 2026. A “normal” CAPE near the long-run median of 16 would imply an index value of about 2,320 on the same earnings denominator. A “normal” CAPE near 25 — the post-1990 median — would imply about 3,625.

None of this is a price target. It is the arithmetic of saying that the current multiple is in the top decile of its 144-year history.

Historical CAPE at major turning points

Date CAPE Context
December 1920 4.78 Post-WWI deflation low
September 1929 32.60 Pre-crash peak (Black Tuesday)
June 1932 5.57 Great Depression trough
January 1966 24.06 End of post-war bull run
August 1982 6.64 Volcker disinflation trough
December 1999 44.19 Dot-com peak (all-time high)
October 2007 27.21 Pre-GFC peak
March 2009 13.32 GFC trough
December 2021 38.58 Post-pandemic / zero-rate peak
June 2026 (current) 41.43 AI-driven mega-cap concentration
Source: multpl.com Shiller PE Ratio, sourced from Robert Shiller's Yale dataset. Long-term mean 17.39, median 16.10. Current value as of June 2026.
Shiller CAPE ratio, 1881 to 2026Line chart of the cyclically adjusted price-earnings ratio for the S&P 500 over more than 140 years, with peaks marked at 1929, 1999 and the present.Shiller CAPE (S&P 500), 1881-2026Long-term mean: 17.41929 (32.6)1999 (44.2)Today (41.4)1900193019601990202001020304050
Source: Shiller, Yale University — ie_data.xls, replicated via multpl.com. Anchor points charted from verified monthly observations.

The all-time high of 44.19 in December 1999 is well documented. Two extremes worth keeping in mind: CAPE bottomed at 4.78 in December 1920 after the post-WWI deflationary depression, and at 6.64 when Paul Volcker's war on inflation broke the long bear market in August 1982. Both troughs preceded multi-decade bull markets.

Why CAPE matters: starting valuation drives long-horizon returns

The most useful empirical fact about CAPE was popularized by AQR's Cliff Asness in a 2012 white paper. Sort every month from 1926 to 2010 into deciles by starting CAPE, then compute the realized real return over the next ten years. The pattern is monotone: cheaper starting CAPE, higher subsequent real return.

Average subsequent 10-year real return by starting Shiller PE decileBar chart showing that higher starting CAPE deciles are followed by lower average real returns over the next decade, falling from roughly 10 percent in the cheapest decile to below 1 percent in the most expensive.Higher starting CAPE -> lower forward 10-year real returnU.S. monthly data, 1926-2010 (Asness, 2012)0%3%6%9%12%10.3%10.4%9.1%8.8%8.0%5.6%5.3%3.9%0.9%0.5%D1D2D3D4D5D6D7D8D9D10Starting CAPE decile (D1 = cheapest, D10 = most expensive)
Source: Cliff Asness, AQR, “An Old Friend: The Stock Market's Shiller PE” (2012). Returns are real, annualized, ten-year forward.
Starting Shiller PE decile Avg 10y real return
1 (cheapest, CAPE <9.6) 10.3%
2 (9.6 – 10.7) 10.4%
3 (10.7 – 11.9) 9.1%
4 (11.9 – 13.8) 8.8%
5 (13.8 – 15.7) 8.0%
6 (15.7 – 17.3) 5.6%
7 (17.3 – 18.9) 5.3%
8 (18.9 – 21.1) 3.9%
9 (21.1 – 25.1) 0.9%
10 (most expensive, >25.1) 0.5%
Source: Cliff Asness, AQR, “An Old Friend: The Stock Market's Shiller PE”, 2012. U.S. monthly data, 1926-2010.

This is not a market-timing signal at a one-year or even a three-year horizon. CAPE's correlation with one-year returns is near zero. It is a long-horizon expected-return tool: when you buy expensive, the average subsequent decade is mediocre.

Where CAPE breaks down

Three criticisms come up over and over again, and each has merit.

1. Accounting regime shifts make ten-year earnings non-comparable. The way GAAP treats goodwill impairments and writedowns has changed materially — most notably after FASB statements 142 and 144 in 2001-02. The huge 2008-09 writedowns dragged the trailing 10-year denominator down for years afterward, mechanically inflating CAPE. Jeremy Siegel has argued that switching to NIPA after-tax corporate profits (a national-income measure) gives a cleaner picture.

2. Interest rates and equity premia. A higher CAPE is partly justified when long-bond yields are low, because the discount rate on future cash flows is lower. The Fed model and earnings-yield-minus-bond-yield comparisons are cruder cousins of this argument. With the 30-year Treasury near 5% in mid-2026, this excuse for a high CAPE is thinner than it was in 2020-21.

3. The composition of the index changes. The S&P 500 in 1980 was 25% energy and industrials by weight; today it is dominated by software and platform businesses with structurally higher margins. Asset-light companies arguably should trade at higher multiples than the historical average. How much higher is the question reasonable people disagree about.

How professionals actually use CAPE

  • As a long-horizon expected-return input, not a timing signal. Vanguard, GMO, and Research Affiliates all publish 7-to-10-year capital market assumptions that lean heavily on starting valuation, and CAPE is the most common anchor.
  • As a portfolio-tilt input: lean toward asset classes whose CAPE is in the lower deciles of their own history (international developed, emerging markets, value vs growth). The within-country, within-asset-class comparison is cleaner than absolute thresholds.
  • As a position-sizing input for tail-risk hedges. When U.S. CAPE is in its top decile, the conditional distribution of forward returns has a fatter left tail. That changes how much it costs to be wrong.
  • Not as a stop-loss. CAPE was above 30 from 1996 through 2000, and the S&P 500 nearly doubled in those four years. The market can stay expensive far longer than skeptics can stay solvent.

Common mistakes

  • Treating CAPE as a sell signal at a fixed level. There is no magic number. A CAPE of 30 in 1929 ended very differently from a CAPE of 30 in 2017.
  • Comparing U.S. CAPE to U.S. CAPE only. The whole point of CAPE is cross-comparison. International CAPEs — published quarterly by Star Capital, Research Affiliates and others — matter at least as much for asset allocation.
  • Forgetting the denominator effect. A recession that crushes earnings will spike CAPE in the short run even though prices are falling. The Asness decile work uses end-of-month CAPE; intra-period spikes don't change the long-run pattern.
  • Mixing real and nominal returns. The Asness decile returns above are real — net of inflation. A 0.5% real return is a low-single-digit nominal return in a 2-3% inflation regime.

Related concepts — what to learn next

  • PEG ratio — growth-adjusted P/E for single stocks.
  • Term premium — the discount-rate input to any DCF.
  • Equity risk premium — the gap between earnings yield (1/CAPE) and the real bond yield. A useful sanity check.
  • Tobin's Q — market value of equity divided by replacement cost of assets. Another long-horizon valuation gauge with a similar message.

Sources & further reading

Disclosure: This article was produced with AI assistance and reviewed before publication. It is for informational purposes only and is not investment advice.

Leave a Comment