Personal Money · Sunday, 30 August 2026

01 Briefing what happened

Two thirds of these funds closed. The scoreboard only shows the ones that lived.

Personal Money 1 min 48 sources

A fund that shuts down stops appearing in the tables people judge funds by. Morningstar counts it as a failure anyway, and the gap between those two habits is most of what an ordinary saver gets wrong about a track record.

66%

of these funds closed

US large-growth funds that were running twenty years ago [1]

13%

survived and beat the index

big US company funds, ten years to June 2026 [1][7]

146

funds shut in 2025

a record, in the year about 1,000 opened [3]

1.75

years a closed fund lasted

the usual life of a 2025 closure [4]

The lead story — what happened

Morningstar's scorecard covers 9,226 US funds holding about $29 trillion, roughly two thirds of the whole US fund market. [1]
It grades a fund on two questions, not one. Did the fund still exist at the end? And did it beat a cheap index fund - a fund with no manager picking, that just holds everything on a list? [1]
Over the ten years to June 2026, 25% of funds run by a manager managed both. [1]
Among the big US company funds it was 13%. The Wall Street Journal reported the same figure off the same data. [7]
The hardest number sits in one corner of the market. Of the funds picking fast-growing big US companies twenty years ago, 66% closed, and fewer than 1% beat their average index rival. [1]
A fund that closes is not in next year's table. That is the whole problem, and it has a name: survivorship bias, the habit of reading a list that had its failures removed. [31]
Morningstar states the rule plainly. A dead fund flunks the first question, because it did not make it to the end. [2]
Watch what dying does to a record. Of 150 US high-yield bond funds alive at the end of 2004, 72 were still there twenty years later. The other 78 closed, after an average of 8.3 years. [2]
Thirty-two of those 78 were ahead of the index on the day they died. But the ones that were ahead lived 2,844 months between them. The ones that were behind lived 4,518. [2]
Funds close for a dull reason. Morningstar puts an exchange-traded fund's fixed running costs at about $250,000 a year, so it needs roughly $33 million in it to break even. [3]
In 2025, 146 funds run by managers and traded on an exchange closed. 114 were wound up and 32 were folded into other funds. It was a record, and so was the roughly 1,000 that launched. [3]
Most of the ones that closed held under $25 million and had been running about 1.75 years. Only six held more than $50 million. [4]

What is pushing on this

Funds being shut High↑

146 exchange-traded funds closed in 2025, alongside 357 ordinary mutual funds [3]

Funds being launched High↑

about 1,000 new ones opened in the same twelve months [3]

How long a fund gets to prove itself Easing↓

the ones closed in 2025 had been running about a year and nine months [4]

Odds of picking one that lasts and wins Easing→

one in four over ten years, and one in eight among big US company funds [1][7]

Who is involved

Morningstar — publishes the scorecard that counts a closed fund as a failure S&P Dow Jones Indices — publishes SPIVA, the rival scorecard, on the same rule Martijn Cremers, Jon Fulkerson and Timothy Riley — three academics who argue the rule is too harsh on the dead Fund companies — decide which funds open, and how long each one gets

How it unfolded

End of 2004 150 US high-yield bond funds are running [2]
By 2024 78 of them have closed, after an average of 8.3 years [2]
2025 about 1,000 exchange-traded funds run by managers open, and 146 close [3]
Feb 2026 Morningstar publishes the count [3]
Aug 2026 the mid-year scorecard puts the ten-year success rate at 25% [1]

Where this points

Morningstar expects more closures: 462 of these funds are both under $50 million and already past the year and a half a manager usually allows. [4]

The rest of the day

26 more stories on this beat.

Each with its own sources. None of these is a link to the story above.

02
One good year did not change it

In 2025 alone, 3,140 funds run by managers were measured. 38% both survived the year and beat their index rival. Over the ten years to the end of 2025, it was about one in five. [47]

Why it matters — A single year flatters or punishes almost at random, which is why the ten-year count is the one that matters.
03
In one corner, 184 of 464 funds died

Morningstar studied 464 funds in five unusual categories that do not follow the main stock and bond markets. Over ten years, 184 of them were wound up or folded into another fund. [48]

Why it matters — Two in five gone, in a group a saver would have picked for safety through variety.
04
Hedge fund closures hit a two-year high

The data firm HFR counted 129 hedge funds wound up in the first quarter of 2026. That is up from 45 the quarter before, and the most since the middle of 2024. In the same three months, 166 new hedge funds opened. [9]

Why it matters — The same shape as the fund tables above: the count published afterwards has the 129 taken out of it.
05
Half of America's stocks lost money

Hendrik Bessembinder of Arizona State University tracked 29,081 US stocks across a hundred years. 52% lost money and 59% did worse than plain cash. The average stock gained 30,621% in total. The middle one lost 6.9%. [10]

Why it matters — An average carried by a handful of huge winners is not what a typical holding did.
06
The winning signal that stopped winning

A 2009 paper said a measure called active share - how far a fund strays from the list it is judged against - predicted which funds would win. AQR re-tested it in 2016 and found no predictive power. A BlackRock test on later data found it pointed the wrong way. [11][33]

Why it matters — A rule found in old data has to survive new data before it means anything.
07
The machine could not pick winners either

Researchers rebuilt a study that used machine learning to choose funds. The advantage vanished. Yearly returns fell by 1.37 to 1.42 percentage points, and none of it held up. The team also found survivorship bias in the original work. [12]

Why it matters — The same defect turned up inside a method built to remove human judgement.
08
Launching a lot and keeping what sticks

Morningstar counted more than 1,000 new exchange-traded funds of every kind in 2025, beating 2024's record of 750. It called the practice a spaghetti cannon: fire enough at the wall and a couple stay up. 276 of the new funds tracked a single company's shares. [5]

Why it matters — It names the business reason the tables lose their failures - the failures were expected all along.
09
Researchers pay to keep the dead funds

The CRSP mutual fund database keeps funds that no longer trade. That way a study of fund returns is not accidentally a study of the ones that lasted. [6]

Why it matters — Somebody had to build a separate product to undo the deletion, which is a fair measure of how much it distorts.
10
Hedge funds are graded more gently still

Managers hand their numbers to databases voluntarily, and often only once the numbers look good. Investopedia calls that backfill bias, and says survivorship makes the same published figures better-looking again. [8]

Why it matters — Two edits to one record, both pushing the same way.
11
Joining the S&P 500 is not the prize it looks

A study of matched pairs of companies found that firms added to the S&P 500 do significantly worse over the long run. That is after the short price jump index funds cause when they buy in. The authors call it a pattern, not a proven cause. [13]

Why it matters — The index's own record is made of arrivals and departures a reader never sees. [32]
12
Savers lost 1.2 points a year to their own timing

Morningstar put the average dollar in US funds at 8.7% a year over the ten years to the end of 2025, against the funds' own 9.9%. The difference comes from when people bought and sold, and adds up to about $3.8 trillion. [14]

Why it matters — Even a surviving fund's published return is not what the people in it actually got.
13
The SAVE repayment plan is gone

SAVE ended on 1 July 2026. [37] Under a settlement, the US Education Department agreed to move the roughly 7 million borrowers still in it onto other plans. It had allowed payments as low as $0 for the lowest earners. [15][16]

Why it matters — It was the cheapest plan for people earning least, and it no longer exists.
14
One new plan, and four fixed lengths

The Repayment Assistance Plan charges between 1% and 10% of income for 30 years, with a $10 floor. [17] A new Tiered Standard plan sets the length by the size of the debt: 10 years under $25,000, rising to 25 years above $100,000. [18][35]

Why it matters — A longer term means a smaller payment each month and more interest paid overall.
15
Grad PLUS loans are shut to new borrowers

More than half a million graduate students took a Grad PLUS loan in 2024-25, and new students no longer can. [19] A lifetime cap of $257,500 now covers most other federal loans. [18][36]

Why it matters — Medical and law students borrowed the largest sums under it.
16
Some forgiveness rules were narrowed too

The administration is rewriting the rules for public service loan forgiveness, which would shut some public workers out of relief they were counting on. [38] Borrowers describe the wider overhaul as confusing and unfinished. [39][34]

Why it matters — People chose those jobs partly on the old rules.
17
Medical debt may return to credit reports

The Consumer Financial Protection Bureau issued an interpretive rule saying federal law overrides the 15 state laws that keep medical debt off credit reports. It is not legally binding, so courts will settle it. As of 2024, 36% of US households had medical debt. [20]

Why it matters — It changes nothing this week and could change a great deal later.
18
North Carolina wiped out 2.5 million people's medical debt

All 99 of the state's hospitals agreed to stop collecting certain debts dating back to 2014. They also agreed to discount care automatically for a family of four earning under $96,000, with no application. [21]

Why it matters — It was done by agreement between a state and its hospitals, not by a law.
19
Elsewhere the state bills mostly failed

Bills to shield patients from medical debt were defeated this year in Indiana, Montana, Nevada, South Dakota and Wyoming, against opposition from the industry. [22] The consumer bureau's own guidance says an unpaid medical bill only reaches a credit report once it is over $500 and more than 365 days late. [41][40]

Why it matters — Protection now depends mostly on which state you happen to fall ill in.
20
Britain now regulates buy now, pay later

From 15 July 2026 the Financial Conduct Authority regulates buy now, pay later credit in the UK. The market grew from £60m in 2017 to more than £13bn in 2024, and use rose from 14% to 25% of adults in a single year. [23]

Why it matters — It was the largest form of everyday borrowing sitting outside the normal credit rules.
21
One in six US adults used it, and it leaves no trace

The Federal Reserve surveyed nearly 13,000 people and found 16% used buy now, pay later in 2025. Most lenders do not report the loans, so paying on time earns no credit and no lender can see what a borrower already owes. 26% of users had paid late at least once. [24]

Why it matters — A debt invisible to every lender is also invisible to the next one deciding what you can afford.
22
The four-payment loan is not the whole product

A Federal Reserve research note in June 2026 mapped what these lenders now sell: alongside the free four-payment split, there are longer loans that do charge interest. [43] The consumer bureau's own study covered six large firms and tracked late fees and unpaid loans from 2019 to 2023. [42]

Why it matters — Two different products share one friendly name, and only one of them is free.
23
The savings rate quoted is not the one most people get

The average US savings account paid 0.38% in August 2026. [25][44] The best widely available accounts paid between 4.2% and 4.5%. [26]

Why it matters — The gap is about eleven times, and it is not a reward for skill - it is a reward for having moved the money.
24
The top rates apply to a sliver of the money

Lists of the highest savings rates are led by accounts that pay their headline number on a thin slice only. One credit union pays 5.25% on the first $500, then less above it. A children's account pays up to 7% on balances up to $1,500. [27]

Why it matters — The rate at the top of the page is real, and almost none of your money can earn it.
25
Higher earners lose a tax break on catch-up saving

From 2026, workers aged 50 and over whose Social Security wages topped $150,000 the previous year must make catch-up contributions after tax, not before. [28] The 401(k) limit rises to $24,500 and the individual retirement account limit to $7,500. [29]

Why it matters — The same money is saved either way; the tax simply moves from later to now.
26
The identical fund, taxed two ways

Brookings researchers compared an exchange-traded fund with an otherwise identical mutual fund. In their worked example the two ended about 25% apart, $61,000 against $49,000. The reason is a tax rule that lets one swap shares instead of selling them, and the benefit lands mostly on higher-income households. [30]

Why it matters — Two funds holding the same shares, and the wrapper decides the tax bill.
27
Buying a first home got slightly easier

US list prices fell 4% in the last quarter of 2025, to an average of $412,800, as rates eased and more homes came up for sale. [45] The typical first-time buyer is now 40 years old. [45][46]

Why it matters — Most of the improvement is the ordinary winter slowdown, not a change in what people can afford.

02 Lesson why it matters

The failures were taken off the list before you read it

A record made only from what survived is not a record of what happened. It is a record of what is left.

The twist

A record built only from what survived is not a record of what happened. It is a record of what is left, and the deletion is invisible because the deleted thing takes its own row away with it.

How it works

Many things start at once
The ones doing badly are quietly shut
Only the survivors stay on the list
The list is averaged, and the average looks strong
A reader treats that average as what happens next

Where you've seen this

Career advice

the people who left the industry are not at the conference telling you what worked

Old buildings

we say they built things to last, having only ever seen the ones still standing

Product reviews

the customers who returned it and moved on rarely come back to write anything

Medicine

a treatment looks better when the trials that failed were never published

The catch

Leaving the dead out is not always an error. Three academics argue funds should get credit for the years they were ahead before closing, and Morningstar disagrees. Both are defensible. The mistake is not knowing which rule made your number.

And the whole of it

Nobody in this story is hiding anything. A fund company shuts a small fund because it costs money to run. A data firm publishes the funds that exist. A saver reads the table in front of them. Every step is reasonable, and the number at the end is still wrong. Each of us is somewhere inside a list like it, reading a record that quietly closed over the parts that did not last.

03 Truth what's really going on

What is really going on

Fund companies open far more funds than they expect to keep, and every one they shut takes its record out of the table the survivors are still being judged in.

Why it works on us — A table looks complete because every row in it is real, and nothing on the page shows you the rows that were taken out.

Who gains

Fund companies — A fund that fails can be closed, and its record leaves the published table with it. Only the survivors keep being advertised. [3][4]
Whoever launches a fund tracking one company's shares — Morningstar says the point is not to beat the market. If a stock catches on, the fees cover the whole batch. It counted 276 such launches in 2025. [5]
Index fund providers — Every scorecard showing 13% of big US company funds surviving and winning is an argument for their product, published by somebody else. [1][7]
Higher-income households — Brookings finds the tax advantage of the exchange-traded wrapper is concentrated among them, and that many households hold only the other kind. [30]
Banks paying the average savings rate — The gap between 0.38% and about 4.2% stays with the bank every year a saver does not move the money. [25][26][44]

Who pays

Anyone picking a fund from a published table — The average in it was worked out after the failures were removed, so the odds it suggests are better than the odds that were actually on offer. [1][2]
People holding the 146 funds that closed in 2025 — A closure hands the money back and makes the holder choose all over again, carrying whatever risk that choice brings. [3][2]
The roughly 7 million borrowers moved off SAVE — The plan with the lowest payments for the lowest earners ended, and they were reassigned to others. [15][16][37]
People with medical debt in the fifteen protected states — Their protection now rests on court cases nobody has finished, after the federal rule that would have removed the debt was struck down in Texas. [20][22]
Buy now, pay later users who pay on time — The loans are mostly not reported, so paying on time earns nothing on a credit file. 26% of users have paid late at least once. [24]

What nobody knows yet

Open questions from across today’s stories — ours included.

01

Whether counting a dead fund as a failure is the right rule.

Morningstar says yes. Three academics say funds should get credit for the years they were ahead before closing. On US stock funds over twenty years, the first rule gives 92% falling short and the second gives 55% of the money. [2]
02

What a saver would actually have earned.

The scorecards count funds, not people. Morningstar's separate study puts the average dollar 1.2 points a year behind the funds themselves, which is a different measurement again. [1][14]
03

Why any particular fund gets shut.

Morningstar can see that funds under about $33 million tend to go, and that the ones closed in 2025 had run about 1.75 years. The decision itself belongs to the fund company and is not published. [3][4]
04

Whether joining the S&P 500 causes the weaker returns that follow.

The authors say their matched pairs show a pattern and an order of events, not a proven cause. It may be that the kind of company added at that moment would have lagged anyway. [13]
05

How much medical debt ends up back on credit reports.

The consumer bureau's rule is not legally binding, fifteen state laws still stand, and no court has ruled. [20]
06

How much people owe on buy now, pay later.

Most lenders do not report the loans, so no credit file holds the total and no other lender can see it. [24]
07

What the new student loan plans cost a borrower over a lifetime.

They started on 1 July 2026 and nobody has finished one. The Repayment Assistance Plan runs for thirty years. [17][18]
08

Whether the ten-year fund figures hold up.

One-year success rates moved 7 points across all categories in twelve months, and 22 points in bonds. A ten-year figure is made of ten of those. [1][47]

04 Hope carry this

Nobody had to count the funds that closed. Someone built a database to keep the failures in, and someone else published the number anyway - which is how a record slowly becomes honest.