Twenty years of deciding what to do with the money
Berkshire has paid no dividend since 1967. Every dollar it earns is reinvested by the same small group of people, which makes the record of those decisions the most direct evidence available about the quality of the management you are buying.
Twenty years of Berkshire's capital allocation show the same decision made repeatedly in different forms: buy a whole business, buy part of one, buy back your own, or wait, and waiting has been chosen more often than any of the others.
- 1 $597 B of operating cash flow since 2006, none of it paid out. $244 B went back into businesses already owned, $147 B bought new ones and $80 B bought Berkshire’s own shares.
- 2 Over the whole period Berkshire was a net seller of quoted shares, by $54.1 B. The famous stock portfolio has been a source of cash across these twenty years, not a use of it, and almost all of that reversal happened after 2022.
- 3 Reported profit is not the signal. Since 2018 an accounting rule has run unrealised portfolio gains through the income statement, producing a $22.8 B loss in 2022, a year the operating businesses earned a record $30.9 B. Read the second line, not the first.
Four places the money can go
“Managing Berkshire is primarily a job of capital allocation, coupled with the selection and retention of outstanding managers to captain our operating subsidiaries.”
Warren Buffett · 2014 Annual Report
Cash arrives from the operating businesses and the investment portfolio. It can be spent in four ways: inside businesses Berkshire already owns, on buying new ones, on buying shares of businesses it will not control, or on buying back its own stock. There is a fifth option, leaving it in Treasury bills, and Berkshire has lately been exercising it.
The chart below puts all four below the line as uses of cash, with operating cash flow drawn across the top as the constraint. Where a bar rises above the line, the cash came back: net equity sales, most dramatically in 2024.
The record, one year at a time
Totals hide the decisions. What matters is what was available in a given year and what was done with it, the Goldman and GE preferreds written in 2008 because Berkshire was the only buyer with cash; the 2010 stock issuance for BNSF that Buffett almost never resorts to; the record buybacks of 2020 and 2021, when his own shares were the best thing on his list.
Pick a year. The bar heights are total cash deployed, so the tall ones are the years something large happened.
Why the equity line swings so hard. Buying quoted shares is the only one of these four uses that can reverse. A railroad cannot be un-built and an acquisition cannot be returned, but a position can be sold, and in 2024 Berkshire sold $134.1 B net, mostly Apple and Bank of America. That single year is larger than the entire twenty-year total of either acquisitions or buybacks.
Two profit lines, one of them meaningless
From 2018 an accounting standard required Berkshire to run unrealised gains and losses on its equity portfolio through the income statement. Because the portfolio is large relative to the earnings, the reported bottom line became a read on the stock market, not on the business.
The result is the gap in the chart below. Buffett has asked shareholders in every letter since to ignore the top line. It is worth seeing exactly how much noise he is asking them to ignore.
The takeaway This is why the valuation on this site is built from normalised operating earnings and not reported profit, and why the essay’s earning-power chapter rebuilds the figure from the parts instead of taking the income statement at face value.
The twenty-year answer
Running the four destinations as cumulative totals settles the question of what Berkshire actually does with money. Capital expenditure wins, and not narrowly: $244 B ploughed back into the railroad, the utilities and the industrial businesses it already owned.
The line that ends lowest is the interesting one. Net purchases of quoted equities finish the period below zero, Berkshire has taken $54.1 B more out of the stock market than it put in since 2006. For a company known for its stock portfolio, that is the most surprising number on the page.
What this implies for the valuation. The upside case assumes $100 B of cash is deployed at 1.5× cost. For scale: across the whole of 2006–2025 Berkshire deployed $147 B into acquisitions. The upside case therefore asks for about 68% of twenty years’ worth of deal-making, and asks for it from here, in a market where the company has lately preferred Treasury bills. That assumption is doing a lot of work, and it is one of the five you can change.
Download the series (.xlsx) Berkshire capital allocation record, with sources and the date it was struck written into the file. This page’s series only; there is no combined download.