Methodology

The honest summary

The US CAPE shown here is the real, measured Shiller series. The headline Global CAPE blends that real US series with developed-ex-US and emerging-market legs anchored to published benchmark levels, so it stays close to professional estimates. Every single-country CAPE is a rough estimate built from free ETF prices, and is not used to compute the headline. Nothing here is investment advice.

What CAPE is

The Cyclically Adjusted Price-to-Earnings ratio (CAPE, also called the Shiller P/E or PE10) divides a market’s inflation-adjusted price by the average of its inflation-adjusted earnings over the prior ten years. Averaging a decade of earnings smooths out the business cycle, giving a steadier read on valuation than a one-year P/E:

CAPE = real price ÷ (10-year average of real earnings)

The core problem

Computing a true CAPE bottom-up for many countries needs ten years of inflation-adjusted index earnings per country. That data sits behind paid, licensed feeds (MSCI, Bloomberg, Barclays/StarCapital). There is no free, long-history, per-country earnings dataset. So a faithful global CAPE cannot be built from free data. Rather than pretend otherwise, we build a defensible estimate and label it as such everywhere it appears.

Data sources (all free, no API key)

We use the split-adjusted close, not the dividend-adjusted close: CAPE is a price ratio, and using a total-return price would inject a spurious upward drift of a few percent per year into the estimate over decades.

Country CAPE estimate (everything except the US)

For each non-US country we anchor a recent CAPE level and roll it through history with the real (CPI-deflated) ETF price, removing a steady real-earnings trend:

realPrice(t) = ETF_close(t) ÷ CPI(t)

CAPE(t) = anchor × [ realPrice(t) ÷ realPrice(t_anchor) ]
                 ÷ (1 + g) ^ (years from t_anchor to t)

• anchor — an approximate recent CAPE level for the country, pinned to 2024-12-01. These are reference values, refined as better free snapshots appear.
• g — assumed long-run growth of real, cycle-smoothed earnings (default 2%/yr). Without it, a market whose real price is flat for a decade would look permanently cheap as its earnings grow.
• Currency & CPI — the ETF prices are USD-denominated, so a raw deflation would let exchange-rate moves masquerade as valuation changes. We instead convert each price back to local currency (using a FRED FX series) and deflate by that country’s own CPI, so the estimate reflects local valuation rather than the dollar. Where a country has no free local CPI (e.g. Taiwan), we fall back to the simpler USD/US-CPI method.

Over short windows this tracks real price (smoothed earnings barely move month to month); over long windows the growth term does real work. It captures directional valuation changes, not a precise earnings-based level.

Aggregating into Global CAPE

The headline Global CAPE is deliberately not a sum of the 16 single-country estimates above. Each of those tracks its ETF price too closely to be reliable in aggregate — in a strong rally the whole ex-US set drifts up at once, pushing the blend well above professional estimates. Instead we combine three robust legs by market-cap weight, as a cap-weighted harmonic mean (total price over total earnings):

Global CAPE(t) = ( Σ wᵢ ) ÷ Σ ( wᵢ ÷ CAPEᵢ(t) )

The two regional legs are pinned to recently-published benchmark levels and rolled only a short way to the present with a broad regional ETF, so the headline stays close to professional global-CAPE estimates (~28) and barely drifts, while the US leg (the majority of the index) is always the real, live series. We refresh the regional benchmarks periodically. The single-country pages remain available as clearly-labelled, bounded rough estimates, but they no longer drive this number.

Data freshness

The canonical free Shiller workbook is updated periodically and can lag the live market by several months, so a US CAPE built purely from it is accurate but not real-time. To keep the headline figure current, we extend the US series to the present month by rolling Shiller’s last measured value forward with the S&P 500 price (the same method used for the country estimates). Those recent months are stored and labelled as estimated, distinct from the genuine measured history, and are flagged on the metric page. This is why our current US CAPE tracks live sources that recompute from today’s price, while everything up to Shiller’s last real observation remains the true series.

Limitations (please read)

1. It is an estimate, not a measurement. No per-country earnings are used except for the US.
2. Anchor sensitivity. Every non-US level scales with its anchor, though the shape over time does not.
3. Single growth assumption. One growth rate per country drives the whole long-run trend.
4. Currency & CPI coverage. Most countries are converted to local currency and deflated by local CPI, but that CPI is annual (interpolated) and a few legs without it stay on the cruder USD/US-CPI method.
5. Price ≠ fundamentals. A price surge with flat earnings reads as a higher CAPE even when it is an earnings story.
6. Shallow history. Non-US legs only reach back to their ETF inception (~1996–2012); only the US extends deep via Shiller.

See the live figures on the metrics pages or the country comparison. Sample (demo) data is always clearly labelled and must never be used for investment decisions.