The Buffett indicator

What the Buffett indicator measures

Take the market value of a country's companies and divide it by what the country produces in a year. In a December 2001 essay in Fortune, written with Carol Loomis, Warren Buffett called this ratio “probably the best single measure of where valuations stand at any given moment” — while noting that it has limits. He compared all publicly traded securities with GNP; most charts today use GDP. Quote: Warren Buffett and Carol Loomis, “Warren Buffett on the Stock Market”, Fortune, December 10, 2001

Our version uses the Federal Reserve's market value of the shares of US nonfinancial companies — listed and private alike, valued at market prices — from its Financial Accounts (Z.1), at the end of each quarter. It leaves out banks and insurers, and the shares of foreign companies. The wider count, all corporate shares, adds financial firms and the foreign shares Americans own (companies whose profits come largely from other economies) and stands at 380% of GDP. The long chart below has both on a switch.

GDP is the Bureau of Economic Analysis's quarterly figure at an annual rate, in current dollars. Both sides of the ratio are in the same dollars, so inflation cancels out. The Fed's share values go back to 1945 and quarterly GDP to 1947, so the ratio starts in Q4 1947; until 1951 there is one reading a year, for December.

Where it stands now

In Q2 2026 the indicator reached 255%: the highest of all 303 readings since 1947, above the previous record of 229% set in Q4 2025. Each dot below is one reading. They pile up where the ratio spent most of its life and thin out to the right; today's is ×3.5 the median.

What is normal? Split all readings into five equal groups and the middle fifth runs from 65% to 81%; the median is 72%. But normal has drifted. Over the last 30 years 99% of readings were above that median, and the median of those years alone is 125%; the ratio has not dipped below the long-run median since Q1 2009.

The long line: highs, lows and a new normal

Its great highs were Q4 1968 (87%), Q1 2000 (163%), Q2 2007 (121%) and Q4 2021 (219%); each was followed by a fall in the ratio of 25% or more. The lowest reading of all came in Q2 1982, at 32%, after years of high inflation and high interest rates.

The line also shows why “cheap” and “expensive” are relative. In the 30 years from 1966, 96% of readings sat in the bottom three fifths; in the last 30 years, 96% sat in the top two.

What it said about the next 10 years

For every quarter from Q4 1947 to Q2 2016 we know what came next: the S&P 500's total return, dividends reinvested, over the following 10 years. Each dot below is one such decade, placed by the reading it started from. The cloud leans the way Buffett would expect — higher readings, lower returns — with a correlation of −0.58 before inflation and −0.41 after it. A real link, but a loose one: the reading accounts for about 17% of the ups and downs in real returns. The rest came from everything else.

After readings in the bottom fifth the median decade returned +11.3% a year after inflation (+15.7% before); not one of those 61 decades lost money. But the fifths are not a neat staircase: after inflation, the weakest median followed the middle fifth. The top fifth is the odd one: its 21 finished decades started in 1998–2000 and 2013–16 and went opposite ways, from −5.9% to +11.8% a year after inflation, and 9 of them lost money. In all, 41 of the 263 decades (16%) lost money after inflation; they started from readings between 64% and 163%.

Today's reading lies outside the chart's known world. The highest start with a finished decade is 163% (Q1 2000), and 255% is ×1.6 that. The 40 readings since Q3 2016 are still waiting for their 10 years: they are the ticks along the bottom, and the dashed line is today.

Where it failed

Same reading on the thermometer, opposite weather. In Q2 1972 and Q1 2009 the indicator read 69% and 69%: practically the same. The decade after the first returned −3.8% a year after inflation; the decade after the second, +14.3%. It happened higher up as well: Q1 1999 at 129% was followed by −5.9% a year, Q2 2016 at 135% by +11.8%.

The ratio sees prices and the size of the economy, but not what money costs or how much of each dollar of output ends up as profit. In the late 1960s and the 1970s, inflation and rising interest rates ate the returns of ordinary-looking readings. In the 2010s, very low rates and wide profit margins carried stocks far above old “expensive” levels — and kept them there.

Not a timing tool

Even when the ratio is right about a decade, it says little about the next year or two. Of the 57 quarters in the top fifth that have a full year after them, the S&P 500 (with dividends) was higher a year later in 44: 77%.

In Q2 1996 the ratio passed the record it had set in Q4 1968. It went on rising for another 3.8 years, and the S&P 500 returned +129% with dividends (+109% after inflation) before the ratio peaked in Q1 2000. The latest old record to fall was the one from Q1 2000, in Q2 2020. Since then the S&P 500 has returned +161% with dividends to the ratio's highest reading so far, and that story is not finished.

Why the level may have moved

None of this means the ratio must return to its old median. Several things change what it measures without changing what stocks are worth:

Each of these can justify a higher level than in 1947; none says how much higher. The ratio is better at telling you that stocks are expensive against their own past than at telling you what they should cost.

Other countries

The World Bank publishes the same ratio for listed companies. Among the 40 largest economies with a reading from 2017 or later, the United States ranks No. 4 at 224% (2025), behind Hong Kong, South Africa and Switzerland. The middle of the list is 67%, and 26 of the 40 economies are below 100%. That figure counts listed companies only, financial ones included, at the end of the year, which is why it differs from the Fed-based 255% above.

Stock-heavy places top such lists for reasons that have little to do with price. Switzerland's exchange lists global companies that earn most of their money abroad, so their value is set against a small home economy: 241% in 2025, 281% at the 2000 peak. Hong Kong is the extreme case, with many mainland Chinese companies listed there: 1,425% in 2025. On the chart its bar runs off the edge.

Not every country reports every year: United Arab Emirates (2024), India (2024), Netherlands (2017), United Kingdom (2022), France (2018), Norway (2019), Israel (2024), Belgium (2018), Russia (2023), Ireland (2018) and Argentina (2022) show older readings, marked with their year. For India the newest value in the source looks implausible and is left out, as on our country pages. A country's own past is a fairer yardstick than its neighbours: the gray bar shows each one's lowest and highest reading since its data began.