The State of Corporate America

The obvious way to ask “how many companies grew free cash flow at 15% a year for 10 years?” is to look at the companies that are still around and count. That answer is wrong, and always too high, because the question has already thrown away everything that failed.

Survivor rate divides by the companies still filing with positive free cash flow at the end. Corrected rate divides by everyone who started — counting the companies that went cash-flow negative, were acquired or stopped filing as the failures they were for anyone holding them. Over 2015–2025 the cohort went from 1,522 companies to 963 survivors, so the two differ substantially.

The chart draws the corrected rate. The survivor rate appears in the tooltip and the table beside it, never on its own.

Companies we cannot judge — no recent successful filing fetch, so an absence might be our gap rather than their failure — are excluded from the cohort entirely rather than counted as deaths (0 here).

Screen for the companies priced above these rates →

Each quarter is a snapshot of every company at once, not a single number. The dark band holds the middle half of the market: a quarter of companies sit above it, a quarter below. The light band holds 80%. The line is the median — the company in the exact middle.

A widening band means the market is spreading out — the same median with more dispersion means winners and losers are pulling apart, which a single line would hide completely.

Why no average or standard deviation. These distributions have long tails and several cross zero. A market P/E includes companies earning almost nothing, so an average is set by a handful of extremes rather than the market: measured across 58 quarters, the raw average P/E ran 3 to 32 times the median and swung between adjacent quarters while the median barely moved. A standard deviation is worse — it is inflated by exactly the outliers it claims to summarise, and because these distributions are not bell-shaped, “average ± 1 SD” does not contain the 68% of companies people assume.

Percentiles have no such problem. They are positions in a sorted list, so one absurd value moves them by one rank at most. For ROIC in the whole market even a trimmed average sits 0.311 interquartile ranges from the median — too far to draw beside it without misleading you, so it is not offered here.

Quarters marked provisional are still filling in — recent filings keep arriving for months, so the newest points are thin rather than meaningful.