Key takeaways
- Vanguard's US defined contribution data for 2023 shows an average 88% equity allocation for participants under 25 and 45% for those aged 70 and over, a 43 point slope.
- In 2005 the same series was hump shaped: 57% under 25, a peak of 71% at ages 35 to 39, and 43% at 70 and over. The slope was 14 points, not 43.
- Nearly all of the change happened at the young end. The under 25 band rose 31 points between 2005 and 2023, while the 70 and over band rose 2 points.
- In UK ISAs, where no default glide path applies, 38.0% of subscribers under 25 chose a stocks and shares ISA in 2022 to 2023, against 35.8% of those aged 55 to 64.
- FCA survey data for 2024 runs the other way: 22% of UK adults aged 18 to 24 held investments, rising to 51% of those aged 75 and over.
The short answer: asset allocation by age slopes down only where a default makes it
You want to know what a portfolio normally looks like at your age, and whether yours is strange. Here's what the observed data says, rather than the rule of thumb.
Where a pension default picks the mix, the equity share falls steeply with age. In Vanguard's US defined contribution plans in 2023, participants under 25 held an average of 88% in equities. Those aged 70 and over held 45%. That is a 43 point slope, and it looks much like the textbook.
Where people choose for themselves, it barely slopes at all. Among UK savers who subscribed to an ISA in the 2022 to 2023 tax year, 38.0% of those under 25 put money into a stocks and shares ISA. At 55 to 64 the figure was 35.8%. The line is close to flat across forty years of working life, then falls to 25.9% at 65 and over.
So asset allocation by age describes what defaults do to people more than it describes what people do. That distinction is the whole piece.
The 401(k) record: hump shaped in 2005, a glide path by 2023
Vanguard publishes a series of observed allocations by age running from 2005 to 2023, in its annual How America Saves report. The chart above plots the 2023 figures across eleven age bands, running from 88% at under 25 down to 45% at 70 and over. Read on its own, it is a tidy confirmation of lifecycle theory.
Then compare 2005. That year the series read 57% under 25 and 64% at 25 to 29. It climbed to a peak of 71% at ages 35 to 39 before falling to 43% at 70 and over. Vanguard describes that shape as "hump-shaped, with younger participants adopting more conservative allocations, middle-aged participants holding the highest equity exposure, and older participants having equity exposure similar to younger participants". The slope from youngest to oldest was 14 points. In 2023 it was 43.
Here is what the comparison exposes. The under 25 band rose 31 points between 2005 and 2023. The 70 and over band rose 2 points. The old end of the curve hardly moved at all. What changed is that young workers stopped sitting in cash and bonds, because something else began choosing for them.
Vanguard says so directly. The 2023 shape is "tied directly to the growing use of target-date funds and managed account advice, both of which provide for declining equity exposure with age". Its March 2026 preview of the next edition, on preliminary year-end 2025 data, reports 82% of participants holding a balanced portfolio. The slope isn't a change of heart. It's a change of default.
Lifecycle theory says the equity share should fall, and it gives a reason
The reason isn't that older people are more nervous. It's human capital, meaning the wages you haven't earned yet.
Someone aged 25 with four decades of salary ahead owns an asset that behaves a little like a bond. It's a long stream of reasonably predictable payments. Set against that, a high equity weight inside the invested pot is less aggressive than it appears. As the wage stream shortens, the bond-like asset shrinks, and the invested pot has to carry more of the risk control itself. That is the argument behind every downward sloping glide path, and it's an argument about the balance sheet rather than about temperament.
Vanguard's February 2024 commentary on age appropriate allocations puts the prescription in numbers. "Many target-date funds have a downward-sloping equity glide path, declining from around 90% equity allocation to about 30%." Hold that against the observed 45% at 70 and over. Even after target date funds reshaped the curve, the oldest band sits 15 points above where a typical glide path ends. The familiar shortcut, the 100 minus age rule, is a cruder version of the same slope, and it lands in a different place again.
In UK ISAs, risk tolerance by age barely moves until 65
UK ISAs are close to a natural experiment, because nobody is defaulted into anything. You open a cash ISA or a stocks and shares ISA, and the decision is yours each year.
HMRC publishes ISA subscriptions split by age and by account type. The left column below is the share of each band's subscribers who put money into a stocks and shares ISA in the 2022 to 2023 tax year. The right column is the FCA's 2024 count of how many adults in each band hold any investments at all.
| Age | Subscribers choosing a stocks and shares ISA, 2022 to 2023 | Adults holding any investments, 2024 |
|---|---|---|
| Under 25 / 18 to 24 | 38.0% | 22% |
| 25 to 34 | 39.7% | 28% |
| 35 to 44 | 36.2% | 29% |
| 45 to 54 | 37.5% | 35% |
| 55 to 64 | 35.8% | 39% |
| 65 and over / 65 to 74 | 25.9% | 44% |
| 75 and over | included above | 51% |
Two things stand out. The left column is flat. Across the four working age bands it sits between 35.8% and 39.7%, and nothing in it resembles a 43 point glide. Only at 65 and over does it break, down to 25.9%.
The right column points the other way entirely. Investment ownership climbs with age, from 22% at 18 to 24 to 51% at 75 and over, a 29 point rise. On the measure the FCA uses, the oldest adults in Britain are more than twice as likely to hold investments as the youngest.
These two columns measure different things, and the difference matters. The HMRC figure is a yes or no about one year's subscription, not a portfolio weight. The FCA figure counts whether someone owns any investments, not how much. Neither is an equity allocation by age in the sense a glide path means it. Both are measures of risk taking, and neither declines the way the theory says it should.
The extremes fell much faster than the average moved
The averages understate what defaults changed. The interesting movement is in the tails.
In Vanguard's plans the share of participants holding no equities at all fell from 13% in 2006 to 3% in 2023. Over the same stretch the share holding nothing but equities fell from 19% to 5%. Those are the two positions that lifecycle theory says are wrong for almost everybody, and both shrank by roughly three quarters.
The residue is concentrated. Among participants who still built their own portfolios in 2023, 8% held no equities and 15% held nothing but equities. So extreme positioning didn't disappear. It retreated into the group that opted out of the default, which is also the group whose allocation shows the least relationship to age.
That matters for reading any average. A plan average of 45% at 70 and over can be most people near 45%, or it can be a mix of people at nothing and people at everything. In the do-it-yourself group it is closer to the second.
Investors kept more of the return in the funds that do the ageing for them
If a default moves allocation, it should show up in returns actually earned. Morningstar's Mind the Gap study measures that gap each year.
Over the ten years to 31 December 2025, the average dollar invested in US funds and exchange traded funds earned 8.7% a year. The funds themselves returned 9.9%. The shortfall of 1.2 percentage points a year is about 12% of the aggregate return, and it comes from when investors bought and sold rather than from what they owned. US stock funds did better: the average dollar there made 12.8% against the funds' 13.3%.
Morningstar's finding across editions is that investors in allocation funds, including target date funds, captured a larger share of their funds' returns than investors in other categories, with US equity the one exception. That is the result worth holding next to the allocation data. The vehicle that produces a downward sloping equity share by age is also the vehicle that gives a saver fewest moments at which to act. What the behavior gap is and how it is measured is a longer argument, taken up separately.
The strongest objection: most of the measured gap is a hindsight artefact
Every number above about investor behaviour rests on dollar weighted returns, and that method has a serious problem. It is worth taking properly, because it cuts at the evidence rather than at the conclusion.
The behaviour gap literature started with a method that compares a market's geometric mean return, which is what a strict buy and hold investor earns, with its dollar weighted return, which accounts for money flowing in and out. Hayley summarises what Dichev found with it: "a 1.3% per annum reduction for equities traded on NYSE and AMEX exchanges (1926 to 2002) and a 5.3% reduction for NASDAQ stocks (1973 to 2002)".
Simon Hayley, then at Cass Business School, showed that this differential mixes two effects. One is the correlation between investor cashflows and future returns, which is genuine mistiming and genuinely costs money. The other is the correlation with past returns, which produces what he calls a hindsight bias and costs nothing. His decomposition of the NYSE and AMEX sample from January 1926 to December 2002 splits the gap "into an annualized -0.21% from adding the return data and -0.95% from adding the distribution data". The geometric mean over that span was 9.87% and the dollar weighted return 8.61%. Hayley cautions that the annualised components sum only approximately, but the ranking is not close: the bias term is more than four times the timing term. Most of the apparent penalty is an artefact of when the market happened to be small.
Hayley's own NASDAQ result is the honest counterweight to his critique. There the geometric mean was 9.6% and the dollar weighted return only 4.2%, and he attributes "the large majority (-4.0%) to bad investor timing, with only -1.0% due to hindsight bias". The main cause he identifies is money arriving during the dotcom boom. So bad timing is real and can be very large, but the size of it depends heavily on the asset and the window. A headline behaviour gap quoted without that decomposition is not a measurement of how badly people choose.
Wealth is doing some of the work that age appears to be doing
The UK picture has a confound worth naming, and it weakens the striking version of the finding.
Investment ownership rises with age partly because money does. In the ONS Wealth and Assets Survey for April 2020 to March 2022, median household wealth rose with the age of the household head, from £15,200 at 16 to 24 to a peak of £502,500 at 65 to 74. HMRC's figures show the same gradient inside ISAs: the average ISA market value was £8,288 for holders under 25 at the end of 2022 to 2023, against £64,386 for holders aged 65 and over.
Investing at all requires a buffer first. If the youngest adults hold fewer investments, some of that is a statement about deposits and rent rather than about risk tolerance vs risk capacity. The FCA's own series makes the point: 90% of UK adults had cash savings in 2024 while 35% had investments. Cash comes first, and it comes first for reasons that have nothing to do with a glide path.
That reading does not rescue the theory, though. Lifecycle theory is a claim about the equity share of invested money, not about who has any. Among the people who do subscribe to an ISA, and who therefore have cleared the buffer, the stocks and shares share still barely moves from 25 to 64.
What this evidence cannot tell you
Start with what each series actually covers. The Vanguard figures are one recordkeeper's defined contribution plans, which is a large sample but not the US population, and they exclude every account a participant holds elsewhere. Somebody at 45% equities inside a 401(k) could be far higher once a taxable brokerage account is counted. Vanguard also sells the target date funds whose growth it credits with reshaping the curve.
The UK numbers are narrower still. ISA subscriptions tell you nothing about workplace pensions, which are themselves defaulted into a glide path. In the ONS survey, private pension wealth made up 35% of household wealth against 14% for net financial wealth, so the pot this piece cannot see is the larger one. If the same analysis were run on UK pension defaults rather than ISAs, the slope would almost certainly reappear, for exactly the reason the 401(k) data shows it.
Both HMRC tables describe the 2022 to 2023 tax year and the FCA survey was fielded in May 2024, so none of this is a reading of current behaviour. The Vanguard series is annual and stops at 2023 in the edition used here. And the whole comparison is cross sectional. A 70 year old in 2023 is a different person from a 25 year old in 2023, with a different history, so asset allocation by age measured this way is not the path any one investor followed.
What would change the conclusion
If the young end of the US curve stopped moving and the old end started. The claim here is that defaults did almost all of the work, and the evidence is that 31 points of change landed on the under 25s and 2 points on the over 70s. A future edition showing the 70 and over band falling towards the 30% a glide path ends at would mean something different was happening, most likely decumulation products rather than accumulation defaults.
If UK pension default data showed a flat line too. That would break the explanation rather than the finding. Defaults would then not be sufficient to produce the slope, and the US pattern would need another cause.
If the behaviour gap were decomposed properly and still looked large for mainstream equities. Hayley's correction left only -0.21% a year of genuine timing cost for NYSE and AMEX stocks. If a replication on a modern sample found most of the gap was timing after all, then allocation funds would be doing more for investors than smoothing a measurement artefact.
The practical question underneath all this isn't what other people your age hold. It is how far your own weights have drifted from the ones you picked, and whether you picked them for a reason you'd still give. LedgerTouch shows the first number continuously. Only you have the second.