A household earning four hundred thousand dollars does not keep its house warmer than a household earning thirty. It may keep a larger house warm. But it does not run the furnace at eighty degrees because it can afford to.
That is the whole theory in one sentence. This year is its clearest test.
What actually went up
Start with the headline. Consumer prices rose about three and a half percent over the year. By the standards of 2022 that is unremarkable.
Now take it apart.
Groceries rose 1.8 percent in September. Food at home should finish the year up about 2.3 percent. That is close to normal. It is genuinely good news.
Gasoline went from $3.14 a gallon to $4.04. That is 29 percent. In August it accounted for more than a third of the entire rise in consumer prices.
Residential electricity hit a record 19.6 cents a kilowatt hour in August. That is about 6.9 percent above last year. The push came from data center demand, which has nothing to do with households at all.
Diesel went from $3.68 to $6.22. That is 69 percent, and it is the freight story told elsewhere on this site.
So the composition is the opposite of comforting. The category that fell quiet is the one a family has some room to manage. The categories that moved are the ones it does not.
Why that distinction decides everything
The Toilet Paper Theory starts from a physical observation. Necessities have a ceiling that income cannot raise and a floor that poverty cannot lower very far.
A rich household does not buy more toilet paper. A poor one goes without.
Energy is the purest case on the board. You can drive fewer miles, but not if the miles are to work. You can set the thermostat lower, but only down to the point where the pipes freeze and the baby gets sick.
Food has more give in it. You can trade down to store brands, which Americans have been doing all year, buy less meat, waste less. The adjustment is miserable and it is possible.
Who pays it
Here is the number this dispatch is named for.
In the Consumer Expenditure Survey, the lowest fifth of American households spends 33.8 percent of its before-tax income on food. Another 18.8 percent goes to utilities, fuels and public services.
That is 52.6 cents of every dollar. Call it fifty-three.
The highest fifth spends 6.4 percent on food and 2.4 percent on utilities. Eight point eight cents.
Six times the share, on precisely the two categories that moved this year.
So take a five percent rise in the cost of keeping a house lit and fed. It costs the top fifth about four tenths of one percent of income. It costs the bottom fifth about two and a half percent.
Same price. Same country. Same month.
How many people this is
The bottom fifth is twenty percent of households by construction, which is about 27 million households out of roughly 137 million.
But the squeeze does not stop at a quintile line, and the better measure is behavior rather than category.
As of February, about 21.5 million American households were behind on their utility bills. That is roughly one in six.
Utilities disconnect around three million households a year. Federal energy assistance reaches about one in five of the households eligible for it. Nearly six million used it last year to avoid a shutoff.
And the average energy burden for a low-income household is 8.6 percent of gross income, against 3.0 percent for everyone else. The Department of Energy calls six percent a high burden and ten percent a severe one.
Why the average hides it
Median household income is about $87,288. The average is $126,664, pulled up by the top.
That gap is the reason the aggregate reads calm. A single national inflation number weights every household the same way, and then gets compared to an income figure that describes almost nobody.
This is the same blind spot the Credit Counterfeit Index admits to in its own methodology. An aggregate cannot see the distribution beneath it. The published reading sits at minus 1.36 and the monetary side of it is at minus 0.01, which says money is not binding.
Money is not binding on average. For 21.5 million households it is binding completely.
What the theory says to do about it
The framework is not neutral here, and the recommendation follows from the physics rather than from politics.
| Approach | What the theory predicts |
|---|---|
| Broad income support | Works, because the constraint is monetary at the household level. A dollar given to a household at the floor is spent entirely, and on the thing in question. |
| Energy bill assistance | Works for the same reason, and is better targeted. The limit is reach rather than design: it currently gets to about one eligible household in five. |
| Price caps on energy | Fails, and this site has been here before. Where the shortage is physical, a cap produces absence rather than affordability. |
| Waiting for the aggregate | Fails by construction. The index that would signal a problem is the one that cannot see this one. |
And the uncomfortable part. The cause of the energy rise is not domestic policy at all. It is a closed strait and a data center build, and neither answers to a vote in Washington.
Which means this is a real shortage producing a monetary emergency in millions of households that had nothing to do with it.
What would prove this wrong
Three tests.
Utility arrears may fall through the winter while energy prices stay high. That would mean households are absorbing the shock, and the floor argument is weaker than claimed.
If the 2026 expenditure survey shows the bottom quintile's food share falling materially, the theory is in trouble. It predicts that share is close to incompressible.
And gasoline may retreat toward $3.20 with no change in Gulf transit. That would mean the price carried a risk premium rather than a supply constraint, and the episode was monetary all along.
Cautions. The spending shares come from the 2023 Consumer Expenditure Survey, the most recent year with full quintile detail. They describe the structure of the burden, not this month’s level. Quintile boundaries move with inflation and the composition of each fifth changes with it. The arrears figure is a trade association estimate rather than a federal statistic. And a household’s income fifth is not permanent. A good deal of movement between them happens within a decade. That matters for reading this as a description of people rather than of a year.