RNZ has been running two stories this month that look unrelated. One is about parents worrying about Jobseeker benefit changes and how hard it is for a teenager to land part-time work. The other is about roughly 3000 hospitality businesses disappearing in a year, with an insolvency rate RNZ puts at 3.3 times the national average for businesses.

Both are really about the same question: who is paying for the young adult who can't get enough hours?

There is a Stats NZ series that quietly tracks the answer. It counts households made up of a couple, their dependent children, and at least one other person. A grandparent. An adult child. A partner of an adult child. A boarder. In the most recent year in the series, 2026, there were 28,600 of those households across the country. That is the lowest count since 2008 (Source: Stats NZ, household-income-by-region).

Look at the shape of the last quarter-century and the drop stands out. In 2001 there were just 13,500 such households. The count climbed through the 2000s, hit 39,000 in 2014, and has been trending down since. Today's figure is 10,400 below that peak, a fall of about 27 percent.

The recent slide is sharp too. The count sat at 35,000 in both 2023 and 2024, dropped to 32,000 in 2025, then to 28,600. That is 6,400 fewer households in three years, roughly 18 percent. Worth saying plainly: that is not the steepest run in the series. Between 2014 and 2017 the same category fell from 39,000 to 30,900, about 21 percent over three years. This is a return to a pattern, not a new one.

The most likely reading is straightforward. This household type is the housing market's shock absorber. When rents bite and young people can't get hours, families absorb the extra person and the count goes up. When the pressure eases, or when the extra person leaves the country, or forms their own household, it goes down. A fall of this size is consistent with fewer families doubling up. It is also consistent with fewer people being around to double up with.

Here is the honest caveat, and it matters more than usual for this series. These are sample survey estimates for a fairly small household category, so the year-to-year wobble is real. The series went 34,300 in 2011, then 31,300, then back up to 36,800 in 2013. Three thousand households can move on sampling variation alone. A three-year direction is more trustworthy than any single year, and the direction here is clearly down.

My view: this is the number to watch if you want to know whether the benefit and job-market debates are landing on families or on the state. Every one of those 28,600 households contains a person who is housed by relatives rather than by a landlord or a lease of their own. That is unpaid, invisible welfare, and nobody budgets for it.

When the count rises, families are picking up the slack. When it falls, someone else is. Neither shows up in a Budget line.