Someone spends four hours a day cooking, cleaning, driving a parent to dialysis and sitting with a child who cannot be left alone. At the end of the year, the official record of what that person did is blank.
GAO has now measured the blank. Unpaid household work was worth "at least $5.6 to 6.0 trillion in 2024," which it puts at "at least 19.2 to 20.2 percent of 2024 GDP." On a typical day, 87.3 percent of people did some of it, for about 3.72 hours.
The number is arresting. It is not the point.
What follows from having no transaction
Gross domestic product counts market transactions. GAO states the exclusion without complaint: unpaid household work "is not included in GDP because it takes place outside of formal market transactions and requires additional resources to calculate."
Fine, as far as GDP goes. But a great many systems are downstream of the same idea — that work is the thing there is a record of a payment for. Retirement credits are calculated from earnings. Contribution histories are built from payroll. Eligibility is assessed against reported income.
GAO, quoting the economists it interviewed, puts the consequence in one clause: valuing unpaid work "can help inform policies that affect caregivers, such as policies intended to improve caregivers' financial security in retirement."
A year of caring for a parent produces no W-2. Not because anyone decided caregiving is worthless, but because the machinery that records value records payments, and no payment occurred. The person's later security is computed from a record that never had a place to put that year.
Three things an uncounted thing cannot do
Once something has no number, it loses three specific capacities. All three are visible in a single week of public records.
1. It cannot be traded off. A budget argument is a comparison. Something with no figure attached does not lose the comparison — it never enters it. GAO's own example is exact: "the decline in economic output during the COVID-19 recession was smaller when this value was included." When paid work stopped, unpaid work did not. A measure that counted only the first recorded a deeper collapse than the country experienced, and every decision taken from that measure was taken from a distorted one.
2. It cannot be claimed against. This is the retirement case above, and it is also the shape of GAO's finding on federal paid administrative leave: OPM "does not know the actual costs of the paid administrative leave used for workforce reduction efforts" because that leave "is reported with other types of general paid administrative leave." A cost with no category of its own cannot be invoiced, appealed, or offset — not because anyone hid it, but because the form has no line for it.
3. It cannot be defended. If a security control is not written into the agreement, nobody can be held to it. GAO found seven of HHS's own cybersecurity control areas missing from the 988 Lifeline's cooperative agreement and three missing from the agreement with the crisis centres. The controls were not rejected. They were simply not written down, which is operationally the same as absent.
Uncounted, uncategorized, unwritten: three words for the same failure.
The remedy is boring on purpose
Notice what the fix is in each case. It is never a change of heart.
- For unpaid work: a satellite account — an official statistic published alongside GDP rather than inside it. GAO does not recommend redefining GDP; it observes that "supplementing" it gives a better understanding.
- For federal leave: GAO asks OPM to "create a new category of paid administrative leave" for workforce reduction efforts. One code.
- For 988: incorporate the missing control areas into the agreements. Drafting.
A category, a code, a clause. None of them is an argument about whether the underlying thing matters. They are the plumbing that lets the argument be had at all — and each one is cheap, which is why the absence is so hard to explain after someone has pointed at it.
What this does not argue
It does not argue that unpaid work should be paid, or that GDP is a bad statistic, or that anyone is owed a particular benefit. Those are real arguments and none of them is settled by a measurement.
It argues something narrower and, I think, harder to disagree with: a thing that has never been counted is not in a weaker bargaining position than a thing that is counted and undervalued. It is in no bargaining position at all. Counting it does not win the argument. It starts one.
Which is why a dry GAO report with a range in it — $5.6 to $6.0 trillion, floor not midpoint, secondary child care excluded — is more consequential than it looks. For the first time, the four hours a day has a figure a budget committee can be asked about.
Source links
- GAO-26-107871 — Unpaid Household Work: Estimated Value and Time Spent on Domestic Tasks and Caregiving
- GAO-26-108477 — the federal leave category that contains two different things
- GAO-26-108836 — the 988 Lifeline's missing control areas
- Bureau of Labor Statistics — American Time Use Survey, the data underneath the estimate