Shiller CAPE (Cyclically Adjusted P/E)

Starter — for Never bought a share

CAPE is the price of the whole stock market divided by its average profit over the past ten years, adjusted for inflation — a long-run value tag instead of a one-year one.

Why not use one year of profit?

A single year can be unusually good or unusually bad. In a recession profits fall, which makes the market look expensive just when it is cheapest; in a boom profits peak, which makes it look cheap just before trouble. Averaging ten years of profit, and adjusting for inflation, smooths that out so the measure is comparing like with like.

What it is telling you

A high CAPE means the market is paying a lot for average historical profit. That has historically been followed by lower average returns over the following ten years, and a low CAPE by higher ones. It has not been a useful guide to what happens next month or next year, and it has been "high" for prolonged periods.

How a beginner should use it

Treat CAPE as context for expectations, not as a signal to buy or sell. When it is historically high, expect moderate returns over a decade and be sceptical of anyone promising a repeat of a boom. When it is historically low, the opposite. Neither reading tells you what to do this week.

What to take away

• CAPE = price ÷ ten-year average inflation-adjusted earnings.
• It smooths the cycle so one extreme year cannot dominate.
• It relates to ten-year returns, not to short-term direction.

Medium — for Invested before, reads the news

CAPE divides the real index level by a decade of real earnings, producing a cycle-neutral valuation whose main predictive power is over long horizons and whose main weakness is that it ignores interest rates.

The construction

Take the index level and convert it into today's purchasing power. Take earnings per index share for the last ten years, convert each year into today's purchasing power using the consumer price index, and average them. Divide. The result is a real price over real earnings, which is why it can be compared across decades with completely different price levels and inflation rates.

What the history shows

The series begins in 1871. It has historically averaged roughly 17. It has been under 5 — in the early 1920s — and above 44 at the peak of the technology bubble in December 1999. High readings such as the late 1920s, the late 1990s and the 2020s have been followed by weak or negative ten-year real returns; low readings such as the late 1970s and early 1980s by strong ones. The relationship is real but loose: the starting CAPE explains a meaningful part of the variation in subsequent decade returns, not all of it.

The earnings-yield inversion

Invert CAPE and you get the cyclically adjusted earnings yield — a percentage return on price. That is the form professionals argue with, because it can be compared with bond yields: if the CAPE earnings yield is 3.5% and the ten-year Treasury yields 4.5%, the equity market is not obviously cheap relative to the risk-free alternative. Most of the "CAPE is too high" debate is really this comparison plus a judgement about what the long-run real rate should be.

The known criticisms

Four get raised repeatedly: accounting standards have changed, so decade-old earnings are not measured the same way as today's; index composition has shifted from capital-heavy industrials to asset-light technology, which changes what a "normal" multiple should be; buybacks mean per-share earnings grow faster than aggregate profits; and the risk-free rate is not constant, so a single historical average CAPE is not a single historical fair value. A careful reading treats CAPE as one input among several.

What to take away

• CAPE is a real price over real ten-year average earnings — inflation-adjusted on both sides.
• It has averaged near 17 since 1871, with extremes below 5 and above 44.
• Its earnings-yield form is the version comparable with bond yields.
• Accounting changes, index composition, buybacks and the rate level are legitimate criticisms — use it as context, not a trigger.

High — for Works with this number already

CAPE is a valuation level with a long-horizon return implication; the professional use is as a prior on expected returns and as a rate-relative comparison, never as a timing device.

What the regression actually says

The documented relationship is between the starting CAPE and the subsequent ten-year real return, and it is negative and statistically meaningful. The important numbers are the residual spread, not the central estimate: at a given starting CAPE, realized decade returns have varied widely, and much of that variation is explained by subsequent changes in the rate environment and by margin behaviour. Practically, a high CAPE shifts the distribution of expected returns down and widens it — it raises the value of a margin of safety and of diversification, not the value of a bearish bet.

The rate-relative framing

Because a business's value is future cash discounted at a rate, the same earnings stream is worth more when rates are low. That is the strongest defence of high CAPE readings in a low-rate era, and the strongest warning when rates normalize. Two ways to formalize it: compare the cyclically adjusted earnings yield with the real risk-free rate plus an equity risk premium, or compute an "excess CAPE yield" (the earnings yield minus the real ten-year yield) as a rough cross-asset comparison. Neither is a model; both are more informative than the raw level.

The composition problem

Comparing a 2020s index to a 1920s index compares a group dominated by asset-light software and platform businesses against one dominated by railroads and heavy industry. Asset-light businesses convert little revenue into book capital and often earn higher margins, which can justify higher multiples — until the moats are tested. A defensible adjustment is to look at CAPE alongside the aggregate profit margin of the index: if margins are structurally above history, the same CAPE means a more expensive market, because the denominator is temporarily elevated relative to its own long-run mean.

How it is used in practice

Three legitimate uses. First, as an input to long-horizon planning: an expected-return prior for a diversified equity allocation, which flows into withdrawal-rate and glide-path decisions. Second, as a risk-management input: high starting valuations argue for wider rebalancing bands and lower assumed returns in a plan, not for concentrated bets. Third, as a cross-market comparison: computing the same measure for other markets and sectors tells you where the long-horizon expected returns are relatively better, which is a portfolio construction question rather than a market call.

The failure mode to avoid

The classic error is treating a valuation level as a signal with a date attached. History contains long periods in which CAPE stayed expensive and returns compounded anyway; being early on a valuation call is functionally the same as being wrong, especially with leverage. The mirror error is dismissing the measure entirely because it did not predict a crash: the claim was never about crashes. It is about what you are paying and what that has meant for long-run returns — an argument for humility about expected returns, not for action.

What to take away

• Use CAPE as a prior on ten-year expected returns and a risk-budgeting input, not as a timing signal.
• Compare the cyclically adjusted earnings yield with the real risk-free rate; the raw level alone is not a valuation verdict.
• Check the denominator: structurally elevated index margins make a given CAPE more expensive than history suggests.
• Being early on a valuation call is being wrong — express a valuation view through position sizing and return assumptions, not through concentrated bearish bets.