The arithmetic is correct. I checked it to the dollar. The problem is that it is a claim about mechanism dressed up as a claim about magnitude — and the evidence on mechanism convicts its own author.
David Bach’s Latte Factor has been in circulation for twenty-five years: skip the daily coffee, invest the money, retire rich. His published figure is that $5 a day at 10 per cent becomes $948,611 over 40 years.
Let us be fair to him twice over before taking it apart.
First, the arithmetic is right. $150 a month at 10 per cent nominal, compounded monthly for 40 years, gives $948,612. It reproduces exactly.
Second, Bach explicitly denies the literal reading. He says it is not about coffee but a metaphor for discretionary spending, and his founding anecdote involves about $10 a day of “latte, muffin, and some snacks.” Most debunks attack a version he does not defend.
So let us attack the version he does.
The scale problem
According to the Bureau of Labor Statistics, the average US household spent $78,535 in 2024. Housing took $26,266 — 33.4 per cent. Transportation, $13,318. Healthcare, $6,197.
Food away from home — every restaurant meal, takeaway, work lunch and school cafeteria for everyone in the household — came to $3,945. That is 5.0 per cent of spending, about $10.81 a day for the whole household.
And all non-alcoholic beverages combined — coffee, tea, soda, juice, bottled water, sports drinks — came to $714 a year. Under one per cent of household spending. $1.96 a day.
Housing, transport, food and healthcare together are 71.2 per cent. The category the advice targets is a rounding error next to the categories it ignores.
The nominal problem
The $948,611 is a pre-inflation number. Deflate it to something like a real return — 7 per cent — and the same forty years of discipline yields about $393,700 in today’s purchasing power. Still a lot. About 41 per cent of the headline.
It also assumes forty uninterrupted years at an all-equity return, with no taxes, no fees, and no bad luck about when the crashes land. Over thirty years instead of forty, it is $339,000.
The mechanism problem, which is the real one
This is the part that should settle it.
Using 41 million observations on the Danish population, researchers found that roughly 85 per cent of people are passive savers who do not respond to price incentives at all. Each dollar of government spending on retirement savings tax subsidies raised total saving by about one cent — nearly all of the rest was wealthy people shifting assets around. Automatic employer contributions, by contrast, passed through to net wealth almost fully.
And a meta-analysis of 201 studies on financial education found interventions explained 0.1 per cent of the variance in financial behaviour, with weaker effects in low-income samples, and that even many hours of instruction had negligible effects twenty months later.
Put those together and the verdict is precise. The deliberate daily act of self-denial — the thing that makes the Latte Factor a good story — is exactly the ingredient that does not work. The boring automatic transfer is the ingredient that does.
Bach has written books called The Automatic Millionaire and The Latte Factor. The research says the first title is right and the second is a distraction.
What is actually squeezing households
Housing. 42.9 million US households were cost-burdened in 2023 — paying more than 30 per cent of income on housing. That includes 22.6 million renters, half of all renters, and 12.1 million renters paying more than half their income. Among renters earning under $30,000, 83 per cent were cost-burdened. No quantity of foregone coffee moves a household from 50 per cent of income on rent to 30.
Medical bills, at volume rather than catastrophe. 17.8 per cent of Americans had medical debt in collections in 2020, totalling around $140 billion — more than all other collections categories combined. The CFPB found that 58 per cent of all third-party collections on US credit reports were medical, with a median of $310. These are ordinary bills, in enormous numbers.
Prices, unevenly. Since 1982–84, medical care prices have risen about 1.95 times as fast as prices generally, and shelter about 1.29 times. Apparel has risen at 0.41 times.
Two things that cut the other way
A piece like this should report the findings that complicate it.
Over the last five years, family health premiums rose 26 per cent against wage growth of 28.6 per cent — slightly slower than wages, in that particular window. And inflation-adjusted published college tuition has fallen over the past decade, down 7 per cent at public four-year institutions, with average net tuition after grant aid roughly half its 2012–13 peak. The “everything outran wages” story is not uniformly true.
Where the honest debate is
The claim that two-thirds of bankruptcies are medical comes from surveys asking filers what contributed. An event-study analysis using credit records found hospital admissions cause fewer than 5 per cent of bankruptcies — while also finding that the earnings lost after hospitalisation exceed the out-of-pocket bills, and that lost income is essentially uninsured.
Both figures are defensible answers to different questions. The causal share is almost certainly well below two-thirds and well above five per cent, and no design currently pins it down. Anyone quoting either number as settled is overreaching.
What to actually do
Automate a transfer on payday, before you see the money. Set it low enough that you will not reverse it. Then raise it when your income rises. That is the entire evidence-based version, and it takes one afternoon.
After that, the leverage is in the big three: what you pay for housing, what you pay for transport, and what your insurance actually covers. Those are hard, slow, sometimes impossible decisions. They are also where the money is.
The Latte Factor survived twenty-five years not because it described household budgets accurately, but because it assigned the problem to the one category small enough for an individual to feel responsible for — and the solution that actually works is too dull to sell a book.
General information, not personalised financial advice. Figures are US national averages and will not describe your household. Consult a qualified adviser about your own circumstances.
