Compounding Math  ·  Part 1 of 2

96% of stocks match cash.
What the other 4% share.

Hendrik Bessembinder pulled every U.S. stock since 1926. Most lost to T-bills. The handful that didn't share three observable characteristics. The cohort rotates.

In 2018, Arizona State finance professor Hendrik Bessembinder published a paper that asked: do stocks outperform Treasury bills? He pulled every common stock that had traded on the NYSE, AMEX, or Nasdaq from July 1926 through December 2016. The sample: 25,967 stocks. He calculated the lifetime buy-and-hold return for each one against the rolling one-month T-bill.

Just 4 percent — 1,092 stocks — accounted for all the net wealth ever created in the U.S. stock market. Five companies (Exxon Mobil, Apple, Microsoft, General Electric, IBM) generated 10 percent of it. Eighty-six companies, less than a third of one percent of the sample, accounted for half. The other 96 percent of stocks, in aggregate, matched cash.

The right tail rotates.

General Electric was once one of the most valuable companies in the world. In Bessembinder's 1926–2016 sample, it sat in the top five wealth creators alongside Exxon, Apple, Microsoft, and IBM. GE has since spun apart into three companies and lost most of the wealth it had created. IBM has been roughly flat for fifteen years. Eastman Kodak, once one of the largest companies in America, no longer exists in any economically meaningful form.

Bessembinder's 2023 follow-up paper, covering 64,000 stocks across 42 markets from 1990 to 2020, makes the rotation visible. The top five global wealth creators bear almost no resemblance to the U.S. top five from the longer sample.

Amazon listed in 1997, year seven of the global sample. Alphabet listed in August 2004. Tencent listed in June 2004. Three of today's five largest wealth creators were either pre-IPO or tiny public companies for half the period that produced them. The right tail is not a fixed group of names. It is a rotating cohort, and the cohort rotates faster as digital business models compound at rates that physical-economy companies cannot match.

25,967 stocks  ·  1926 — 2016

4%

of stocks accounted for all net wealth created. The other 96% matched Treasury bills.

What the 4% share.

The academic literature on what separates right-tail compounders from the rest, including Bessembinder's own follow-up work and Morgan Stanley's Counterpoint Global research, converges on three observable properties.

Sustained high return on invested capital. Right-tail names earn 20 percent or more on incremental capital deployed, year after year, for decades. Not occasionally. Persistently. Apple's ROIC has averaged above 30 percent for over a decade. Microsoft's has run above 25 percent since 2014. A company earning 8 percent on incremental capital cannot compound into the right tail no matter how large its absolute capital base. The arithmetic does not allow it.

Reinvestment runway at the same rate. High ROIC matters only if there is somewhere to reinvest at that rate. Right-tail names operate in markets large enough, with penetration low enough, that incremental capital keeps earning the same high return. When the runway ends, even a high-ROIC company stops compounding. This is the GE and IBM problem. The historical ROIC was excellent. The reinvestment opportunities ran out before the company did.

Capital allocation discipline. Right-tail names tend to have low rates of acquisition activity, especially financial M&A. They reinvest organically, return cash through buybacks at conservative valuations, and avoid the corporate-finance entropy that destroys long-tenured compounders. Berkshire is the canonical case. Costco and Visa are recent ones.

These three properties are observable in any 10-K. ROIC and reinvestment runway are calculable from financial statements. Capital allocation discipline is visible across a five-year window of cash flow statements and proxy filings. A reader can screen on all three.

Why owning them is hard.

Identifying a right-tail name is one problem. Holding it is a different problem entirely. Morgan Stanley's Counterpoint Global, building on Bessembinder's data, documented Apple's path to becoming the largest wealth creator in history. The stock generated nearly $2.7 trillion in shareholder wealth from its 1980 IPO through the end of 2022. It also suffered three drawdowns of 70 percent or more along the way.

A 74 percent decline from May 1983 to August 1985. An 80 percent decline from February 1992 to December 1997. A 79 percent decline from March 2000 to March 2003. Three separate occasions when a long-term Apple holder watched four-fifths of their position evaporate. Each time, the data still supported Apple as a right-tail name. Almost no one held through all three.

This is the cost of concentrated active investing. A small book of right-tail candidates is the structurally correct active strategy by Bessembinder's data. It is also the least diversified bet investors can make. Investors who cannot stomach repeated 70 percent drawdowns in individual positions should hold the index and skip the active book entirely.

Berkshire is one of the names in the right tail. Bessembinder's data put Berkshire in the top contributors to U.S. shareholder wealth creation since 1926. Saturday in Omaha is, in part, a meeting about whether the right-tail position can be maintained under Greg Abel.

Have a good Thursday.

— Marques