Key takeaways
- No published paper derives the 60/40 portfolio. The oldest thing you can point at is Vanguard's Wellington Fund, running since 1929 on a 60% to 70% equity mandate.
- UK defined benefit schemes held 61.1% equities and 28.3% bonds in 2006. By 31 March 2025 the weighted average was 15.1% equities and 70.6% bonds.
- In calendar 2022 Vanguard's 20/80 fund fell 13.93% and its 80/20 fund fell 17.09%. Four times the equity exposure cost 3.16 percentage points more.
- Over the ten years to 31 December 2025 Vanguard's 20/80, 40/60, 60/40 and 80/20 funds returned 4.06%, 6.12%, 8.09% and 10.03% a year.
- The US 10-year Treasury yield was 0.52% on 4 August 2020 and 4.74% on 21 August 2026, so the same 40% bond sleeve now starts somewhere else entirely.
Nobody derived the 60/40 portfolio, and that is the honest answer
Where did 60/40 come from? You'll see the ratio quoted everywhere as though it were settled, and almost never sourced.
The short version: there is no founding paper. Nobody optimised their way to 60 and 40. What you can point at is a balanced fund launched in 1929, a 1952 theory that produces a curve rather than a number, and decades of pension practice. The pension funds have since left. UK defined benefit schemes held 61.1% of their assets in equities in 2006. By 31 March 2025 that figure was 15.1%.
None of which makes 60/40 a bad allocation. It makes it a default: a number you inherited rather than chose. What follows traces what the ratio actually rests on, and then tests the two things it needs to be true.
A 1929 balanced fund is the oldest thing you can point at
Vanguard's Wellington Fund is the usual origin story, and it's the only part of the story with a paper trail. The fund's annual report filed with the SEC on 3 February 2026 says of its advisor that "the firm has advised the fund since its inception in 1929."
The mandate today is still a band. The summary prospectus dated 27 March 2026 says the fund invests "60%-70% of its assets in dividend-paying and, to a lesser extent, non-dividend-paying common stocks of established large companies", and puts "the remaining 30% to 40% of its assets mainly in fixed income securities".
So a fund that predates Markowitz's 1952 paper by more than two decades is still run on a 60-to-70 equity band. That is worth knowing, and it's worth being precise about what it proves. What the filings say is that the fund has existed since 1929 and that this is its mandate now. They do not say what the band was in 1929, they give no reason for the band, and the band isn't 60/40 anyway: it's a range with 60 at the bottom of it. A mandate is a product design that survived, not a derivation.
Markowitz gave you a frontier, not a ratio
The other thing people gesture at is Harry Markowitz. In his Nobel Lecture of 7 December 1990 he described how portfolio theory arrived: "the natural approach for an economics student was to imagine the investor selecting a point from the set of Pareto optimal expected return, variance of return combinations, now known as the efficient frontier."
Read that slowly. The theory hands you a curve of the best available risk-and-return combinations, and then you pick a point on it. Which point depends on how much variance you are prepared to carry. The maths does not choose for you.
It also cannot run without your forecasts. Markowitz records that in 1956 he published the "critical line algorithm" for tracing out the efficient frontier "given estimates of expected returns, variances and covariances, for any number of securities subject to various kinds of constraints". Estimates. Move your expected return for bonds by a point and the frontier moves, and every allocation on it moves with it. Mean-variance analysis is a machine for turning your beliefs into weights. It has no way to produce 60/40 unless you feed it beliefs that produce 60/40.
The institutions that ran 60/40 have almost entirely left it
Here is the part that usually gets left out. The Pension Protection Fund publishes an annual census of UK defined benefit schemes, and its asset allocation table runs back to 2006. In its 2006 dataset the weighted average scheme held 61.1% equities and 28.3% bonds, with the rest in property, cash and insurance policies. The equity weight sat within a point of the 60 in 60/40.
By The Purple Book 2025, the same table reads 15.1% equities and 70.6% bonds. That is a fall of 46 percentage points in the equity weight in under two decades, and the book records annuities at "a record high at just under 13 per cent". One caveat on timing: the book says that nearly all the asset allocations in the 2025 dataset "were at a date on or since 31 March 2023", so these are recent scheme returns rather than a single snapshot.
The counter-argument deserves its due. A DB scheme isn't a household. It is matching a stream of promised payments, most of its members have stopped accruing benefits, and its bond figure covers liability-driven investment that behaves like bonds without being a return-seeking holding. A household has no liability to match, so none of this shows 60/40 is wrong for an individual.
What it does dispose of is the appeal to authority. "The professionals run 60/40" was roughly true in 2006. It isn't true now.
The bond leg's job depends entirely on the yield it starts from
A 60/40 portfolio makes a promise about the 40. The bond sleeve is meant to pay you something, and to hold up when equities fall. Both of those depend on the yield it starts from, and that number has moved much further than the ratio ever has.
The US Treasury's own daily par yield curve puts the 10-year at 0.52% on 4 August 2020. At the close of 2021 it was 1.52%. On 21 August 2026 it was 4.74%.
Those are three different assets wearing one label. A 40% sleeve yielding 1.52% has almost no income to offer and a great deal of duration to lose if yields rise, which is what happened next. The same sleeve at 4.74% starts with 3.22 points more yield, and that yield is the raw material for both the income and the cushion. Nothing about the number 40 tells you which of those you hold.
Bonds stopped hedging equities in mid-2021
The second assumption matters more. A stock bond allocation only diversifies if the two legs don't fall together.
The Bank for International Settlements examined this directly. In a box published in its Quarterly Review on 4 December 2023, Marco Jacopo Lombardi and Vladyslav Sushko write that "the correlation between US equity and government bond returns switched sign in mid-2021", and that "one has to go back to the 1980s and the early 1990s to find a prolonged period with positive correlations."
Their mechanism is about what markets are listening to. When inflation is low and steady, traders read economic news mainly for what it says about growth. Weak growth hurts shares and helps bonds, because rate cuts become more likely, so the two move in opposite directions. When inflation is high and volatile, the inflation print takes over. A hot number hurts bonds directly, and hurts shares by removing the prospect of cuts. Both fall together.
So the negative equity bond correlation that the whole case for a balanced portfolio rests on isn't a law of nature. On the BIS reading it is a feature of a low-inflation policy regime, and it lasted roughly two decades before 2021. That is also the finding behind what worked and what didn't in diversification in a crisis.
2022 is the cleanest test anyone has run, and the bond leg failed it
There is a natural experiment sitting in the filings. Vanguard runs four LifeStrategy funds with one advisor, one set of underlying Vanguard funds and four different stock bond allocations: approximately 20/80, 40/60, 60/40 and 80/20. One year, one process, four mixes.
In calendar 2022 they returned -13.93%, -14.99%, -16.00% and -17.09%.
Look at the ends of that row. Just 3.16 percentage points separate the most defensive allocation from the most aggressive one. The 20/80 fund holds a quarter of the equity exposure of the 80/20 fund, and it still lost close to 14%. That is what a positive correlation does to a balanced portfolio: the ratio stops being the thing that determines the outcome.
The worst quarters make the point more sharply still. Across 2018 to 2025, the calendar years the current prospectus charts, the 20/80 fund's worst quarter was -6.93%, in the three months to 30 June 2022. That was a bond quarter. The 40/60 fund's worst was -9.05%, in the same quarter. The 60/40 and 80/20 funds had their worst quarters in the three months to 31 March 2020, at -12.89% and -17.60%, when equities crashed. The most conservative of the four had its worst moment of those years delivered by the part that was supposed to be the safe part. That is a different failure mode from the one described in equity allocation tested in a crash, where the equity sleeve does the damage.
Allocation against outcome: a straight line with no kink at 60/40
Put the four funds side by side over the same ten years, to 31 December 2025, and the shape of the decision becomes visible.
| Stocks/bonds | 10-year return a year | Calendar 2022 | Worst quarter, 2018 to 2025 |
|---|---|---|---|
| 20/80 | 4.06% | -13.93% | -6.93% (Q2 2022) |
| 40/60 | 6.12% | -14.99% | -9.05% (Q2 2022) |
| 60/40 | 8.09% | -16.00% | -12.89% (Q1 2020) |
| 80/20 | 10.03% | -17.09% | -17.60% (Q1 2020) |
Every 20 points of equity added roughly 2 points of annualised return, and the worst quarter deepens by 10.67 points from one end of the table to the other. The chart above plots that first column. It is close to a straight line, and nothing at all happens at 60. There is no elbow, no optimum, no point where the trade-off suddenly improves.
That is what calling 60/40 a default means. Somebody had to choose a point on a smooth line, and 60 is a round number sitting near the middle of it. Had the line bent at 63/37, we would all be quoting 63/37 instead.
The strongest case for 60/40 is about drawdowns, not returns
The best defence of the ratio doesn't come from anyone selling it. It comes from the longest dataset in the field.
Elroy Dimson, Paul Marsh and Mike Staunton have assembled annual returns for 35 markets going back to 1900. Their 2026 Yearbook, covering 126 years of market history, reports this: "Since 1900, equities and bonds have on several occasions lost more than 70% in real terms. yet a 60:40 equity:bond blend has never declined more than 50%."
That is a serious finding, and it survives everything above. Over the same 126 years US equities returned 6.6% a year in real terms and US bonds 1.6%. The blend gave up real return, 5.0 points a year of it at the extreme, and bought a shallower floor with the proceeds. It is a coherent trade. A household that cares more about the depth of the hole than the height of the ceiling can take it and be reasoning correctly, and the loss recovery arithmetic is why depth matters so much.
Two things sit alongside it. The first is that "never" describes one sample of 126 years, not the future. The second comes from the same yearbook: by the end of 2025, it reports, "correlations between DM and EM and between equities and bonds have also risen". The mechanism that produced the shallower floor is the one the BIS says changed sign.
What this evidence cannot tell you
Start with the sample. Four funds from one manager over ten years is not the space of possible allocations, and the LifeStrategy range is a particular portfolio: it hedges its foreign bonds back to the dollar and carries a large foreign equity sleeve. A UK investor's 60/40, in sterling and holding gilts, is a different animal, and this piece has not tested it.
The decade to December 2025 contains one equity crash, one inflation shock and a long bull market. Reorder those and the table reorders with them. Ten years is a sample, not a law, and 2022 is one observation of a positive correlation rather than proof that it persists.
The Treasury yields are US yields. The Purple Book measures schemes matching liabilities, which is a different problem from the one a household has. The Wellington mandate is a band, and an actively managed one, so it tells you what a manager was permitted to do rather than what the fund held on any given day. And the 50% figure from the Yearbook is a maximum over one history: it says the blend has not done worse than that, not that it cannot.
What would change the conclusion
If the equity bond correlation turns negative again and stays there for a decade, the 40 does its old job, and the case for the ratio gets its mechanism back. The BIS box is dated December 2023. The sign of that correlation is the single number this whole argument turns on, and it is observable.
If starting yields fall back toward the 0.52% of August 2020, the bond sleeve becomes what it was in 2022 again: an asset with a small known return and a lot of duration risk attached. The arithmetic that made that year painful has not gone anywhere.
And if somebody produces the derivation, a paper that reaches 60 and 40 from stated assumptions rather than from custom, then this piece is wrong about the history. Nobody has yet.
A default isn't a mistake. It is an answer to a question you were never asked. The question is why 60 rather than 55 or 70, and the honest form of it is about how deep a fall you can hold through, which is a fact about you rather than about the frontier. What is measurable is the other half: portfolio drift, how far your weights have wandered from the ones you set, and LedgerTouch shows that continuously. Whichever ratio you picked, the number worth checking is whether you still own it.