Three to six months of expenses. It is the most repeated number in personal finance. It was set in 1996 by asking 156 financial advisers what they thought — and the largest study of household bank data ever conducted calls it “not empirically grounded.”
Every guide to saving money contains the same instruction: build an emergency fund of three to six months’ expenses. It is stated the way you state a law of physics — no source, no derivation, no acknowledgement that anyone might reasonably disagree.
So where did it come from? The answer turns out to be genuinely interesting, and it changes what you should do with your next hundred pounds.
The rule was set by polling advisers, not by studying households
The closest thing to a founding document is a 1996 paper in Financial Services Review by Greninger, Hampton, Kitt and Achacoso, which set out to establish benchmarks for household financial health.
Its method was a Delphi study: three rounds of questionnaires sent to a panel of 156 experts — 85 practising financial planners and 71 educators — asking what benchmarks they believed were correct. On emergency savings the panel converged on liquid assets worth 250 to 300 per cent of monthly expenses, with a median of 300 per cent.
Read that again, because it is the whole story. No household was observed. No hardship was measured. No outcome was tracked. The study measured what advisers already believed, and the belief became the benchmark.
It was not even a settled belief. A 1995 paper by Chang and Huston noted that planners recommended “anywhere from two or three months to six months, and in some cases a year’s worth” — and, tellingly, attributed that guidance to nobody, because by 1995 there was no originating source to cite. The earlier academic work it reviewed, going back to 1968, built measurement frameworks rather than targets.
The three-to-six-month rule has no inventor. It is a professional convention that hardened into a number.
What the data actually says a household needs
In 2019 the JPMorgan Chase Institute did what the 1996 panel could not: it watched households. Six million families, a 75-month balanced panel of anonymised transaction data, stress-tested against the shock the emergency fund is meant to absorb: a month when income dips and expenses spike at the same time.
Two findings. First, that simultaneous shock arrives roughly once every 5.5 years — not constantly. Second, weathering it takes about 6.2 weeks of take-home income. Around $2,600 for the lowest income quintile, $5,000 for the middle, $14,400 for the highest.
Six weeks. Not six months. And the report is blunt about the conventional advice, stating that three to six months of expenses “is unrealistic for many families” and “is not empirically grounded.”
The independent data agrees on the scale of shock people actually face. Pew’s survey of American family finances found 60 per cent of households had experienced a financial shock in the previous twelve months — and that the median cost of the most expensive one was $2,000. For most households that is well under a single month of expenses.
The first few hundred does most of the work
This is the finding that should change behaviour, and it is the one almost never reported.
The Urban Institute tracked 28,000 families through the Survey of Income and Program Participation and mapped hardship against savings. Moving from under $250 in savings to just $250–$749 was associated with:
- Eviction falling from 3.2 per cent to 0.7 per cent
- Missed housing payments falling from 21.1 to 15.2 per cent
- Missed utility payments falling from 24.5 to 18.5 per cent
Then the curve flattens hard. Missed housing payments fall to 10.8 per cent at $2,000–$4,999, to 6.1 per cent at $5,000–$19,999, and to 3.6 per cent above $20,000. Getting from nothing to a few hundred pounds cuts eviction risk by roughly three quarters. Getting from $5,000 to $20,000 — a far larger sum, and years more saving — moves it by 2.5 percentage points.
The protective value of an emergency fund is front-loaded to a degree the standard advice completely obscures.
Why quoting six months may actively backfire
Here is where this stops being trivia. The Consumer Financial Protection Bureau’s own review of savings interventions — which, notably, contains no discussion of the three-to-six-month benchmark at all — records that in matched-savings programmes, people treat the threshold they are given as the target they aim at. The number you hand someone becomes the number they measure themselves against.
Now consider who is being handed this one. In the Federal Reserve’s 2025 survey of nearly 13,000 adults, 55 per cent had savings covering three months of expenses. Thirty per cent could not reach three months by any means at all — not by borrowing, not by selling something. Among adults earning under $25,000, just 21 per cent met the three-month mark.
And this is not only a low-income problem. Federal Reserve analysis of the Survey of Consumer Finances found 72 per cent of all families lacked six months of expenses in liquid savings — and that even in the top income quartile, only 68 per cent met the three-month benchmark. A target missed by a third of the richest quarter of households is not functioning as guidance.
Told to build something unreachable, the rational response is not to start. The rule may be quietly discouraging exactly the people for whom the first $500 would do the most.
“Months of expenses” assumes a number that does not hold still
There is a further problem with the instruction, which is that it asks you to multiply a figure most households do not have.
JPMorgan Chase Institute found median-income families’ expenses swung by nearly $1,300, or 29 per cent, month to month. In an earlier analysis of 2.5 million account holders, 41 per cent saw month-to-month income swings above 30 per cent, and 60 per cent saw consumption swings that large.
Pew found that among households whose income dropped year on year, the median fall was 49 per cent. Those same volatile households held the least savings — a median of $1,550, against $5,500 for households with stable income.
“Six months of expenses” sounds precise. Built on a denominator that moves by a third, it is not.
What to do instead
The evidence supports a different shape of advice — a sequence rather than a target:
- Get to a few hundred first, and treat that as a real milestone. This is where the steepest reduction in hardship sits. It is not a consolation prize on the way to the real goal.
- Aim next at roughly six weeks of take-home pay, not months of expenses. It is the figure the transaction data supports, and income is a number you can actually look up.
- Keep it genuinely liquid. A buffer that takes three days and a penalty to reach is not a buffer.
- Do not treat saving and repaying debt as strictly either/or. Twenty-seven per cent of households deliberately do both, and modelling suggests this is less irrational than it looks, because some expenses cannot be met with a credit card. The CFPB’s review also found participants who built savings did not offset it by taking on more debt.
- Then keep going, knowing the returns are flattening. More buffer is better. It is just not better at the rate the rule implies.
What this evidence cannot tell you
Almost none of it is causal. Households with $500 saved differ from households with $50 in ways — income stability, family support, health — that independently predict eviction. The drop from 3.2 to 0.7 per cent is a real correlation; how much of it savings causes is not measured.
Nobody has run the experiment the rule implies: randomly assigning households to a three-month buffer versus six and tracking what happens. The six-week figure is a simulation-based stress test, not an outcome trial. And the surveys disagree with each other by roughly ten points depending on whether you ask people what they would do or look at their balance sheet.
What can be said is that the rule was set by asking advisers rather than watching households, that the first few hundred pounds does a disproportionate share of the protecting, and that quoting an unreachable target to the people at greatest risk is a poor way to get them to save the amount that would actually have helped.
This article is general information, not personalised financial advice. It does not account for your circumstances, debts or obligations. Figures are drawn from US federal surveys and US household data; the underlying patterns travel, the specific dollar amounts may not. Consult a qualified adviser before making financial decisions.
