This is the plain-language edition. For the full evidence chain, every number's exact caliber, and all verification verdicts, see the deep dive. This is an evidence review, not investment advice.
"Passive investing has passed half the market, and price discovery is broken." That sentence has two halves. The first is true only on one particular ruler. The second is, as of today, neither supported nor refuted by the evidence.
What is most worth recording is not who won. It is that almost every signature number both sides cite falls apart when you take it back to the original document. We sent three independent people to try to refute each of 32 key claims, then added two more to each of the four most important numbers — one hunting for contradicting evidence, one with the power to veto outright. The result: one vetoed, three restricted, and zero that can be used at the strength people use them.
Start with the simplest and most abused question: how much of the market is passive?
If you measure money inside funds: passive did pass active among US long-term funds at the end of 2023 — $13.29 trillion against $13.23 trillion, a gap of $60 billion, or 0.2% of the total, basically inside classification error. Switch to the fund industry association's own definition and the crossing happened in 2024. By the end of 2025 it is roughly 53% to 55%. On this ruler, "passive passed half" is true.
If you measure stock market value: passive funds' share of total US stock market value is 19% by the industry association's count, 23% on Vanguard's wider method, and 14% by a Federal Reserve staff estimate for 2020. All three describe themselves as understatements, and none is anywhere near half.
If you measure trading: Vanguard says index funds' own trading is 1.2% of US stock market volume; Michael Green's components add to roughly 64%. Both appear in the same news article. This is not a measurement dispute — they count entirely different things. Vanguard counts only funds buying and selling their own holdings; Green includes market-maker trading "attributable to" passive flows, which is 36 percentage points of his total and has no traceable source anywhere.
Is the true passive share higher than the fund-based number? Yes — and that direction has four independent teams behind it. The most convincing bypasses all estimation: researchers with transaction-level records from the Colombian exchange, including the identity of each trader, watched who actually bought and sold around index changes and found that institutions that are nominally active but tracking an index generate rebalancing purchases 1.7 times larger than declared index funds. Even BlackRock's 2017 white paper — written specifically to argue that indexing is not very big — broke out global indexing at 17.5%, of which about 10 percentage points was institutions doing indexing internally, versus 7.4% visible in funds. A firm with every incentive to report a small number still produced more than double.
But how much higher, nobody can pin: independent measurements span 17.5% to 41%. The most-cited "33.5%" comes from one paper whose own table lists five figures, from 13.31% to 33.5% — and about the most conservative one, the authors state plainly that it cannot be statistically distinguished from the 16% fund-based number.
In 2019, Michael Burry told Bloomberg that passive investing "has removed price discovery from the equity markets," comparing it to the synthetic CDOs of the financial crisis. His concrete evidence was liquidity: that day, 1,049 stocks in the Russell 2000 traded under $5 million, and "the theater keeps getting more crowded, but the exit door is the same as it always was." He also wrote: "However, I just don't know what the timeline will be."
What happened: the S&P 500 went from 2,937 that day to 7,798 by August 2026 — up 165% in seven years, before dividends. The stampede he described did not materialize in any of the three stress tests since (March 2020, August 2024, April 2025). His fund shrank from $340 million to $155 million and deregistered in November 2025. He now writes a paid newsletter whose main theme is an AI bubble — hyperscaler depreciation accounting, not index funds. Across 79 published posts, not one restates the 2019 argument.
In February 2024, David Einhorn said the sharpest line of this cycle: "I view the markets as fundamentally broken... almost the best way to get your stock to go up is to start by being overvalued."
From the same interview: "it's actually much more exciting now... there's way fewer people competing with us... the opportunity is actually probably as good or better than it's ever been." So he shifted to buying companies at four or five times earnings that return 15-20% of their value in cash.
Those two statements do not contradict each other — and they are exactly what the classic economic model predicts: fewer informed traders means the remaining ones earn more. By November of that year, his phrasing had softened to "a bit of a broken market."
Michael Green is the theorist of this position and the only one with a timetable. In January 2026 he said on a podcast that markets fail at around 83% passive, that "we're about 54% now, and picked up 4 points last year," and therefore "about five years out the world comes to an end."
Checking each piece:
In fairness: Green's directional call has been right. He is not a crash-caller — his claim is "until this stops, you can't beat the index, and markets keep going up," which is what happened. That makes his thesis hard to falsify, and it means a rising market is not a refutation. In August 2026 he left his firm to found an asset manager whose strategy is trading the price pressure of passive flows — the warner turned the argument into a product.
British fund manager Terry Smith wrote in his July 2026 letter that the market is "dominated by so-called passive or index funds" and the AI boom, and that the ending would be "badly." The cost is real: his fund fell 2.9% in the first half while its comparator rose 11.2%, a 14.1 point shortfall; he raised half-year portfolio turnover to 51.8% (from 3.2% for all of the prior year) and explicitly built momentum into his process.
But his numbers have problems too: the passive share went from "more than 50%" to "more than 60%" in eight months, neither sourced; one sentence claims "in 2007-08, the S&P fell 57% in five months" — it was 17 months; and the five biggest detractors he lists are all stock-specific choices.
The authority the warning camp quotes most stands on the other side. Jack Bogle's 2018 article warned about ownership concentration — that a handful of institutions would end up with voting control of nearly every large company — and contains not a word about price discovery or bubbles. It closes by quoting someone approvingly: "A ban on indexing would clearly not be a good idea. I can only say, 'Amen.'"
His 2017 line, "If everybody indexed, the only word you could use is chaos, catastrophe," is always quoted alone. The next sentence: "Now, what are the chances of everybody indexing? It's zero." Later in the same answer: "So, I'm not concerned about it."
He also suggested indexing could exceed 75% of the market without becoming dangerous. Where did 75% come from? Cliff Asness later recorded the answer in a footnote: he pressed Bogle on it and was told, "I made it up."
Vanguard's 1.2% comes from the biggest beneficiary of debunking. The most-cited scoreboard of active underperformance is published by a company that sells indices. And the flagship material for "bond ETFs provided price discovery in the crisis" is a BlackRock slide deck — routinely mistaken for a regulator's finding because it is hosted on the SEC's website.
Three voices refuse to read from their camp's script. Asness (AQR) concedes markets have become less efficient but judges indexing "a relatively small part of the story," with social media and gamified trading the bigger suspects. Lamont (Acadian) is an active manager — the party the bubble narrative would let off the hook — and writes the sharpest rebuttal: "Since passive investors mostly do not trade, it is mostly true that they do not impact market prices... no trade, no crime." And GMO, famously bearish, judges passive's impact "overblown but not negligible," and shrinking, because the pension-system shift that drove the flows is ending.
This is the part that needs patience, because "price discovery" has at least six meanings in the research, and each gives a different answer.
The cleanest experiment says no. Researchers exploited the fact that a stock's passive ownership jumps when it crosses a threshold in an annual index reshuffle — an accident of ranking, not of company quality. The result was a precisely estimated zero: how far prices stray from a random walk, how mispriced they are on known anomalies, how much they drift after earnings — none of it moved. Over the same period (2007 to 2016), index-fund ownership went from 2% to 11% of market value while the efficiency measures slightly improved.
But the same paper's other half says something did change. Information production fell — after joining a small-cap index, downloads of company filings and analyst report counts both declined. Volatility, correlation with the index, market beta, turnover, and short interest all rose significantly. People who cite this paper for "passive is harmless" tend to skip the second half.
The strongest "damage" evidence was vetoed. A widely cited study reported that the higher a stock's passive ownership, the less of an earnings report was already in the price before publication, quantified as: 15% more passive ownership means about a quarter less pre-announcement information. Our audit seat vetoed that number:
The mechanism layer does hold up, though, and it is multi-source. Set that number aside and the direction — more passive ownership, less company-specific information in prices — is confirmed by three independent teams using three completely different methods, one of which never looks at an earnings window at all.
But the same facts have an opposite reading. Another researcher independently documented the same phenomenon — price drift ahead of earnings weakening — and concluded markets got more efficient: earnings-day prices now move essentially in one step, and post-earnings drift, one of the most famous market anomalies of the past few decades, has been dead since about 2006. The same facts, opposite signs, and the available measures cannot tell them apart.
The mechanical impact of passive buying on a stock price should grow as passive grows. The opposite happened.
When a stock joins the S&P 500 it used to jump immediately: 7.4% on average in the 1990s. By 2010-2020 that had converged to 0.83%, statistically insignificant — while funds tracking the S&P 500 grew from nearly zero to about 7% of US market value. The demand shock more than doubled and the price response vanished.
Who was on the other side? Researchers checked: not active mutual funds (their net buying is about zero), but hedge funds, pension funds and endowments. The market grew its own capacity to absorb a predictable shock.
Then came three echoes in 2025: Coinbase rose 24% the session after its inclusion was announced, Robinhood 15.8%, AppLovin 11.6%. All three are outside the paper's sample and none has a peer-reviewed re-estimation. They do not overturn "the effect converged to zero," but they put a boundary marker on "disappeared."
This is the most popular current version: passive buys in proportion to market value, so money flows disproportionately to the largest companies and inflates the Magnificent Seven.
The strongest supporting evidence is a 2025 paper: quarters with larger index-fund inflows saw the biggest companies outperform the index more, in a pattern that rises with size. It was restricted for hard reasons:
The mechanism does have independent support — including a real-world natural experiment. The Bank of Japan spent years buying stock ETFs, an enormous buyer that does not look at price and buys by index; research confirms it raised constituent prices without near-term reversal while lowering price informativeness and liquidity. But Japan proves that huge price-insensitive buying moves prices, not that the effect is stronger for the largest companies — its cross-sectional variation comes from purchase intensity, not company size.
A closing fact: the Federal Reserve, which published the canonical paper on passive-investing risk, mentions "passive," "index fund" and "ETF" exactly zero times in its May 2026 Financial Stability Report.
If passive really created abundant mispricing, the remaining active managers should be feeding well. The data say otherwise.
Twenty-year underperformance runs 91-94% by fund count. Across 2001-2024, a majority of large-cap funds beat the index in only three years — 2005, 2007, 2009 — and none since 2010. And there is no visible correspondence with the rise of passive: the line is essentially flat.
Two corrections have to travel with those figures. First, half of that 88% fifteen-year underperformance rate is "did not survive" — a fund that closes or merges is scored as underperforming, and whether it was actually lagging while alive is precisely what is in dispute. Second, measured by money rather than fund count, the 20-year annualized gap shrinks from 1.96 points to 1.25. In May 2026 the active industry published a paper, funded by its trade association, arguing that after three methodological adjustments the 20-year figure drops from 92% to 55%. The rebuttal came quickly: nearly a fifth of the improvement in one category was replication error.
There is no referee for this fight. But what is in dispute is the magnitude, not the direction: no one's numbers show active outperformance as the norm.
If any part of this debate rests on solid evidence, it is not price discovery — it is concentration of power. And the most concentrated layer is not the funds; it is the companies that build the indices.
The SEC wrote in 2022 that three firms hold more than two-thirds of the index market, and that constructing an index "leave[s] room for significant discretion," sometimes "without publicly disclosing their index methodologies." Academic research finds that more than a third of ETF expense ratios is paid to index providers as licensing fees. More striking: the same research finds this market cannot fix itself — a new low-fee index provider entering barely moves equilibrium fees.
And the regulatory ending: that request for comment never became a rule and was formally withdrawn from the agenda in April 2025. US index providers remain outside investment-adviser regulation, while their European and British counterparts must register.
The Big Three (BlackRock, Vanguard, State Street) hold roughly 20% of S&P 500 companies on average and cast about 25% of votes. The famous projection is 34% within a decade — but it extrapolates a fixed annual rate, and their own data show the share flat across 2019-2021 (21.5%, 21.1%, 21.9%). In 2026 all three split their stewardship teams in two, which some read as an effort to stay under the 5% filing threshold; but in practice, the two teams often vote the same way — the legal form has split, the voting behavior has not.
Ordered from most to least confident, each with its test.
1. "Passive passed half" holds only on the fund-asset ruler; the market-cap ruler is 19-23%. The direction that the true share exceeds the fund-based count has four independent teams behind it (measurements spanning 17.5% to 41%), but no specific figure is pinned by a second team. Test: the annual industry updates; whether anyone independently replicates the method that infers passive from rebalancing-day volume (nobody has).
2. Active large-cap funds underperform over the long run (20 years: 91-94% by count, 55-78% by money), and this did not improve as passive grew after 2010. Test: the annual scoreboards; whether the trade-funded paper clears peer review.
3. [contested] Whether passive made individual-stock demand less "elastic." One structural model says yes (down about 11% — though the same paper gives 6.3% on another measure), while direct measurement of how much index trading actually moves prices reports that ability falling from 6.75 to 0.37. Worse, the 11% rests entirely on a statistical instrument the model generates itself: remove it and the finding that "competition offsets two-thirds" goes to zero. Test: re-estimate without the same model.
4. [mechanism multi-source / magnitude vetoed] The company-specific information content of prices falls with passive ownership — three independent teams, three measures — but the claim that pre-announcement information therefore fell about a quarter is vetoed. Test: extend those three measures to 2021-2026; test directly whether "less anticipation" or "more information released with the report" is doing the work.
5. [contested] The same "information arrives later" facts can be read as markets getting worse or better. One researcher documents that earnings-day pricing is now nearly one-step and post-earnings drift has disappeared, and reads it as more efficient. Test: find a measure that separates the two readings.
6. [mechanism multi-source / magnitude not usable] Price-insensitive index buying does raise prices (the Bank of Japan experiment is the cleanest); but "29% for the giants over 25 years" is not usable — the authors' own model gives 0-4%, and the direction failed out of sample twice (2022 and 2026). Test: re-estimate the size gradient on post-2021 data.
7. The S&P 500 inclusion effect converged from 7.4% in the 1990s to nearly zero, then three 2025 additions produced double-digit moves (24%, 15.8%, 11.6%). Test: any re-estimation covering the period after 2021; the next large addition.
8. [contested] Whether inclusion causes stocks to move more with the index: a 2016 study says no, a 2025 study with a stronger design says yes. Test: extend the event study to current data.
9. Official autopsies of five "passive fragility" events attribute none of them to passive ownership or index-fund flows. Test: the next stress event's official report.
10. There is directional consensus on a volatility channel, but the calibrated magnitude is under one percentage point. Test: the volatility structure once passive passes 25% of market value.
11. Both the 1.2% and the 64% for "passive share of trading" are constructed, and the largest component (36%) has no source. Test: either side publishing a reconciliation.
12. The Big Three concentration numbers are real but have plateaued; the 2026 split is so far only a legal form. Test: the first data on whether the two teams vote differently.
13. Index providers are the most solidly evidenced concentration in the whole file, and regulators formally withdrew in April 2025. Test: whether the SEC reopens it.
14. The warning side's most testable threshold conflicts with its own paper (83% versus 65%/90%), and its demonstration arithmetic is off by nearly half against its own input. Test: the formal threshold once that paper clears review.
Ask which ruler first. The same "passive share" is 55% of fund assets, 19% of stock market value, and either 1.2% or 64% of trading. Almost every sentence of the form "now that passive is past half, therefore..." has quietly swapped rulers mid-argument.
Both sides' signature numbers took comparable damage. The warning side: the most testable threshold is 18 points off its own paper, the demonstration arithmetic is off by nearly half, and the most famous witness (Bogle) was warning about something else entirely and explicitly defended index funds. The defense is no cleaner: the most-cited study of ETFs and volatility has an abstract reading "a one standard deviation increase in ETF ownership is associated with an increase of 16% in daily stock volatility," while its own body text says that 16% is 16% of a standard deviation — 20 basis points of daily volatility, about 10% of the mean, and only for S&P 500 stocks. That caliber error originated in the authors' own abstract, and the published version quietly dropped it. Several other key papers contradict their own tables, and the active-underperformance gap shrinks by a third when measured in money.
Nearly everything that survives scrutiny lives at the mechanism layer, not the magnitude layer. Company-specific information in prices does fall with passive ownership; huge price-insensitive buying does move prices; shadow passive really is far larger than the fund count. But no independent evidence answers the question "how much distortion has this actually caused" — and the entire force of the bubble thesis comes from magnitude.
This field's clock stopped in the 2010s. Every empirical study's independent variable sits in a world where passive is under 16% of market value; today it is approaching 20%. The "harmless" findings testify about a level of passive that no longer exists; the "harmful" prophecies are still waiting for their sample.
The solid problem isn't about prices; it's about power. Three firms build two-thirds of the world's indices, take a third of ETF fees, exercise enormous discretion, sit outside investment-adviser regulation — and the regulator formally dropped the question in April 2025. That part is not disputed, has primary documents, and almost nobody is arguing about it.
Related research on this site: the same caliber physical — one percentage differing 40-fold across three denominators — appears in A Physical for "95% of AI Pilots Fail"; for the gap between capital-market narratives and primary numbers, see Is This AI Capex Boom Another 1999?; for the structure in which both sides' signature numbers take equal damage, see Screen Time and Teen Mental Health.