Quick Note: Households vs Population as Supply Denominator

My friend Luca Gattoni-Celli just published a good blog post that looks at a series of graphs and explains how they are illustrative of a housing shortage. He is writing about the US, but an equally revealing set of graphs could be found for the UK. (John Burn-Murdoch, the Crown Prince of Interesting Graphs, has made a good start.) Today, I wanted to pick up on something that Luca mentions right at the beginning of his piece:

This first chart, showing new privately-owned housing units started (red) and completed (blue) divided by total households, might be the ultimate documentation of the U.S. housing shortage. Dividing by households rather than population is important because one household inhabits one housing unit (at least in theory) and U.S. household size is shrinking over time as fewer individuals marry and have children.

This distinction between households and population in the denominator is an interesting one, and it compresses quite a bit of information about the dynamics of the housing market.

‘Household’ has a precise statistical meaning, one that differs from folk assumptions. The UK’s Office for National Statistics defines a household as:

one person living alone, or a group of people (not necessarily related) living at the same address who share cooking facilities and share a living room, sitting room or dining area.

(The US Census Bureau offers roughly the same definition.) Crucially, people living in a household aren’t necessarily related to one another. It is not an indivisible family unit, nor is it merely a collection of individuals, like the population measure.

My main point is that the definition of a household is more flexible than the definition of a family. Economic pressures can force unrelated individuals to live together, masking the true demand for housing: young adults living with roommates in house shares when they’d prefer their own space; couples living with a lodger when they’d rather have a child; children living with their parents for longer than desired due to high housing costs. Liam Halligan calls these ‘concealed households’: between the 2011 and 2021 Censuses, the number of families of adult children living with their parents in England and Wales increased by 13.6%.

This ‘doubling up’ phenomenon artificially reduces the number of households when housing is unaffordable, even as the desire for separate living spaces remains high – and it distorts the housing-to-population ratio accordingly. High housing costs can therefore produce a situation where the dwellings-to-population ratio stays the same but the dwellings-to-households ratio gets bigger.

This effect is particularly pronounced in high-demand markets, like the UK. Vacancy rates in the UK are extremely low, suggesting that new dwellings are absorbed almost instantly by pent-up demand for household formation. In other words, we are so supply-constrained that a new unit of housing, on the margin, results in the formation of a new household.

This sensitivity to economics works in the other direction too: as societies get richer, people want to consume more housing. They want larger homes, or to have a shorter commute, or live closer to greater amenities. So as the population gets wealthier we might expect to see a decrease in household size and an increase in household quantity, which, of course, we have – until recently.

All of this to say: household formation is a function of housing costs. If housing costs change, then individuals change their living arrangements, and this affects how many households there are.

So there’s a sense in which Luca is right: dwellings-to-households is more responsive to market conditions, better reflecting changes in living arrangements due to housing costs. Seeing a change in dwellings-to-households while dwellings-to-population remains static, or vice versa, can indicate changes to household size due to economic pressures, demographic shifts, urbanisation, or a mismatch of housing stock and demand; all important things to know about. 

But dwellings-to-population is useful too, precisely because it is less sensitive to the marginal effects of housing costs. In high-cost areas, dwellings-to-households could suggest a sufficient housing supply when what’s really happening is that people are doubling up because of affordability issues. If my current flatmate moves into a newbuild, the dwellings-to-households ratio hasn’t changed: both dwellings and households have increased by one. But the quality of our housing situations have both improved. Dwellings-to-population is also easier to compare across countries, since patterns of household formation are skewed by cultural and demographic factors, as well as brute affordability.

What should a YIMBY care about? YIMBYism is, at root, about increasing affordable housing supply; the YIMBY should first aim to push up dwellings-to-population, as this directly reflects more housing per person. If dwellings-to-households doesn’t increase along with dwellings-to-population, then it’s a red flag that there are affordability issues preventing household formation. But we need to look at both as to not be confounded by either.